The Effects of Roads and Culverts on Stream and Stream-side
Salamander Communities in Eastern West Virginia
Ryan Lee Ward
Thesis submitted to the
Davis College of Agriculture, Forestry, and Consumer Sciences
at West Virginia University
in partial fulfillment of the requirements for a degree of
Master of Science
in
Wildlife and Fisheries Resource Management
James T. Anderson, Ph.D., Major Advisor
J. Todd Petty, Ph.D., Co-Advisor
Division of Forestry
P.O. Box 6125
Morgantown, WV 26506
J. Steven Kite, Ph.D.
Department of Geology and Geography
West Virginia University
P.O. Box 6300
Morgantown, WV 26506
2005
Keywords: stream salamanders, culverts, passage, roads, habitat fragmentation, northern twolined salamanders, Eurycea bislineata, Appalachian seal salamanders, Desmognathus monticol,
northern spring salamanders, Gyrinophilus porphyriticus, mountain dusky salamanders,
Desmognathus ochrophaeus
Abstract
The Effects of Roads and Culverts on Stream and Stream-side
Salamander Communities in Eastern West Virginia
Ryan Lee Ward
Roads and culverts have many effects on stream salamanders. We examined 120 culvert
crossings in the Shavers Fork and Dry Fork watersheds in eastern West Virginia. Complete
barriers to salamander passage occurred at 55.0% of the sites visited and partial barriers at
34.2%. Analyses showed that 20.6% of the total stream length in the Dry Fork watershed and
18.4% in the Shavers Fork watershed were isolated by barriers. We conducted salamander
sampling at 16 streams and used Akaike’s Information Criterion to select the best a priori models
explaining salamander diversity, richness, and abundance. Roads benefited disturbance tolerant
species while negatively affecting other species. Mitigation efforts should focus on lessening
impacts on salamander habitat by preventing negative effects to streams such as sedimentation,
pollution, excessive disruption to the riparian zone, and barriers to movement.
Acknowledgements
I thank the following individuals for field work assistance: Jason Clingerman, Jared
Gregory, Anthony Grubb, Thomas “The Ultimate” Hardeman, Zina Hense, Patrick Kish, Ross
Kite, Seth Lemley, Zachary Liller, Michael Nicholas, Ira Poplar-Jeffers, and Joshua White. To
the culvert crew of 2003, may we never forget gourmet hotdogs, hillbilly campfires, and a coan.
To the field crew of 2004, for the rest of our lives may we wake up, eat our doughnuts, and go to
work.
I extend my thanks to George Seidel for help with statistical analysis. I also thank my
committee for their advice and support, especially James T. Anderson, my major advisor, for the
opportunity to conduct this research and further my education. I thank the West Virginia
Division of Highways, West Virginia University Division of Forestry, and the West Virginia
University Agricultural and Forestry Experiment Station for funding and support to conduct this
research.
Finally, I thank my family, especially my wife, Heather, for their support of me through
hard times and their encouragement to keep going.
iii
Table of Contents
Chapter I ...................................................................................................................................1
Overview of effects of roads and culverts on salamanders and descriptions of study area ..........2
Introduction and Justification ..................................................................................................2
Study Area ..............................................................................................................................5
Shavers Fork .........................................................................................................................6
Dry Fork ...............................................................................................................................7
Gandy Creek and upper Dry Fork........................................................................................7
Laurel Fork .........................................................................................................................8
Glady Fork ..........................................................................................................................8
Red Creek ...........................................................................................................................9
Lower Dry Fork ................................................................................................................10
Literature review ...................................................................................................................10
Salamanders and Habitat .....................................................................................................10
Habitat Fragmentation.........................................................................................................13
Culverts and Salamander Passage........................................................................................14
Fish Interactions..................................................................................................................16
Exploitative Competition...................................................................................................17
Demographic Effects.........................................................................................................17
Spatial Distribution ...........................................................................................................20
Conclusions...........................................................................................................................21
Literature Cited .....................................................................................................................22
Tables. ..................................................................................................................................33
Figures ..................................................................................................................................34
Chapter II................................................................................................................................66
Culvert Effects on Stream and Stream-side Salamander Habitats in the Dry Fork and Lower
Shavers Fork Watershed in West Virginia .................................................................................67
Abstract.................................................................................................................................67
Introduction...........................................................................................................................68
Study Area ............................................................................................................................72
Methods ................................................................................................................................73
GIS .....................................................................................................................................73
Culvert Surveys...................................................................................................................73
Data Analysis ......................................................................................................................74
Results ..................................................................................................................................76
Culvert Surveys...................................................................................................................76
Habitat Fragmentation.........................................................................................................76
Outlet Hang.........................................................................................................................77
Continuous Substrate...........................................................................................................77
Discussion.............................................................................................................................77
Outlet Hang.........................................................................................................................78
Continuous Substrate...........................................................................................................80
Conclusions and Management Implications...........................................................................83
Acknowledgements ...............................................................................................................86
Literature Cited .....................................................................................................................86
Tables ...................................................................................................................................92
iv
Figures ..................................................................................................................................95
Chapter III ............................................................................................................................102
Effects of Road Crossings on Stream and Stream-side Salamander Diversity, Richness, and
Abundance..............................................................................................................................103
Abstract...............................................................................................................................103
Study area ...........................................................................................................................108
Methods ..............................................................................................................................108
Salamander sampling ........................................................................................................109
Habitat assessment ............................................................................................................110
Data analysis .....................................................................................................................112
Leaf Litter Bag Sampling ..................................................................................................114
Results ................................................................................................................................114
Transect Sampling.............................................................................................................114
Leaf Litter Bag Sampling ..................................................................................................118
Discussion...........................................................................................................................118
Salamander Diversity and Richness...................................................................................118
Salamander Abundance .....................................................................................................120
Leaf Litter Bag Sampling ..................................................................................................122
Conclusions and Management Implications..........................................................................123
Acknowledgements .............................................................................................................125
Literature Cited ...................................................................................................................125
Tables .................................................................................................................................132
Figures ................................................................................................................................146
Chapter IV ............................................................................................................................149
Conclusions and Management Implications for Roads and Stream Salamanders....................150
Abstract...............................................................................................................................150
Introduction.........................................................................................................................150
Study Area and Methods .....................................................................................................152
Results ................................................................................................................................153
Conclusions.........................................................................................................................154
Management Implications ...................................................................................................155
Future Research ..................................................................................................................158
Mitigation Opportunities .....................................................................................................159
Literature Cited ...................................................................................................................161
Figures ................................................................................................................................165
Appendices ............................................................................................................................166
v
Chapter I: List of Tables
Table 1: Salamander species of interest in the Shavers Fork and Dry Fork
watersheds, according to Green and Pauley (1987)………………………………... 33
Chapter I: List of Figures
Figure 1. Shavers Fork and Dry Fork watersheds located in Randolph and Tucker
County in eastern WestVirginia……………………………………………………. 34
Figure 2. The Shavers Fork watershed in Randolph and Tucker County, West
Virginia…………………………………………………………………………….. 35
Figure 3. Land cover type in the Shavers Fork watershed………………………… 36
Figure 4. Culvert sites surveyed in the lower Shavers Fork watershed…………… 37
Figure 5. Tributary to Gandy Creek at site 420 shows a typical small stream in
the Dry Fork watershed…………………………………………………………….. 38
Figure 6. Gandy Creek and upper Dry Fork watershed located in Randolph
County, West Virginia……………………………………………………………... 39
Figure 7. Land cover type in Gandy Creek and upper Dry Fork watershed………. 40
Figure 8. Culvert sites surveyed in the Gandy Creek and upper Dry Fork
watersheds………………………………………………………………………….. 41
Figure 9. Salamander sites sampled in the Gandy Creek and upper Dry Fork
watershed…………………………………………………………………………... 42
Figure 10. Laurel Fork watershed located in Randolph and Tucker County, West
Virginia…………………………………………………………………………….. 43
Figure 11. Land cover type in the Laurel Fork watershed………………………… 44
Figure 12. Glady Fork watershed located in Randolph and Tucker County, West
Virginia…………………………………………………………………………….. 45
Figure 13. Land cover type in the Glady Fork watershed…………………………. 46
Figure 14. Salamander sites sampled in the Glady Fork watershed………………. 47
Figure 15. Red Creek watershed located in Tucker County, West Virginia………. 48
vi
Figure 16. Land cover type in the Red Creek watershed………………………….. 49
Figure 17. Lower Dry Fork watershed located in Tucker County, West Virginia... 50
Figure 18. Land cover type in lower Dry Fork watershed………………………… 51
Figure 19. Culvert sites surveyed in the lower Dry Fork watershed……………… 52
Figure 20. Salamander sites sampled in the lower Dry Fork watershed…………... 53
Figure 21. Upstream of site 420 showing typical shade levels on streams in the
study area away from vegetation breaks caused by the presence of roads………… 54
Figure 22. Culverts were often undersized such as this culvert at site 201.
Undersized culverts lead to ponding of water and aggradation at higher flows…… 55
Figure 23. Limestone boulders placed in the stream channel downstream from
site 201. Channelization is common at road crossings……………………………. 56
Figure 24. Common types of culverts……………………………………………... 57
Figure 25. Corrugated steel pipe is a common construction material used for
culverts……………………………………………………………………………... 58
Figure 26. Concrete is a common construction material used for culverts………... 59
Figure 27. High flow velocity through culverts commonly prevents the passage
of aquatic organisms……………………………………………………………….. 60
Figure 28. Lack of sufficient water depth commonly prevents the passage of
aquatic organisms....................................................................................................... 61
Figure 29. Lack of a pool at the culvert outlet commonly prevents the passage of
aquatic organisms…………………………………………………………………... 62
Figure 30. Excessive outlet hang commonly prevents passage of aquatic
organisms…………………………………………………………………………... 63
Figure 31. The retention of bedload inside culverts creates varied flow velocities
and promotes the passage of aquatic organisms…………………………………… 64
Figure 32. Brook trout are commonly found in small streams within the study
area and are predators of salamanders……………………………………………... 65
vii
Chapter II: List of Tables
Table 1. Lengths of stream affected by barrier culverts located on state roads in
the Dry Fork and Shavers Fork watersheds, West Virginia, 2003…………………. 92
Table 2. Results of analyses on retention of streambed substrate performed on 53
single, circular culverts constructed of corrugated steel pipe in the Dry Fork and
Shavers Fork watersheds, West Virginia, 2003……………………………………. 93
Table 3. Results of analyses on retention of streambed substrate performed on 29
single, pipe arch culverts constructed of corrugated steel pipe in the Dry Fork and
Shavers Fork watersheds, West Virginia, 2003……………………………………. 94
Chapter II: List of Figures
Figure 1. Map of study area in the lower Shavers Fork and Dry Fork watersheds,
West Virginia, 2003. Circles indicate state culverts where surveys were
conducted…………………………………………………………………………... 95
Figure 2. Decision tree used to determine barrier status of culverts for stream
salamanders in the lower Shavers Fork and Dry Fork watersheds. Culverts were
classified as complete barriers (n = 66), partial barriers (n = 41), and nonbarriers
(n = 13)……………………………………………………………………………... 96
Figure 3. Frequency of barrier categories for salamanders in the lower Shavers
Fork (n = 52) and Dry Fork (n = 68) watersheds, West Virginia, 2003…………… 97
Figure 4. Linear regression showing the relationship between outlet hang height
and stream gradient for 116 single barrel culverts in the Dry Fork and Shavers
Fork watersheds, West Virginia, 2003……………………………………………... 98
Figure 5. Linear regression showing the relationship between outlet hang height
and culvert slope for 116 single barrel culverts in the Dry Fork and Shavers Fork
watersheds, West Virginia, 2003…………………………………………………... 99
Figure 6. Linear regression showing the relationship between outlet hang height
and culvert length for 116 single barrel culverts in the Dry Fork and Shavers Fork
watersheds, West Virginia, 2003…………………………………………………... 100
Figure 7. Graph showing frequency distribution of areas with continuous
substrate at culvert sites for 120 culverts in the Dry Fork and Shavers Fork
watersheds, West Virginia, 2003…………………………………………………... 101
viii
Chapter III: List of Tables
Table 1. Habitat variables used in models for salamander diversity and
abundance for the Dry Fork, Gandy Creek, and Glady Fork watersheds, West
Virginia, 2004……………………………………………………………………… 132
Table 2. Salamander captures along transects in the Dry Fork, Gandy Creek, and
Glady Fork watersheds, West Virginia, 2004……………………………………… 133
Table 3. Linear regression models of Simpson’s index of diversity and species
richness for salamander communities in streams in the Dry Fork, Gandy Creek,
and Glady Fork watersheds, West Virginia, 2004. The best approximating
models, selected using Akaike’s Information Criterion corrected for small sample
size, are in bold…………………………………………………………………….. 134
Table 4. Logistic models with a negative binomial distribution explaining
salamander abundance on a reach scale in the Dry Fork, Gandy Creek, and Glady
Fork watersheds, West Virginia, 2004. The best approximating models, selected
using Akaike’s Information Criterion corrected for small sample size, are in bold.. 136
Table 5. Logistic models with a negative binomial distribution explaining
salamander abundance on a stream scale in the Dry Fork, Gandy Creek, and
Glady Fork watersheds, West Virginia, 2004. The best approximating models,
selected using Akaike’s Information Criterion corrected for small sample size, are
in bold……………………………………………………………………………… 140
Table 6. Salamander captures for streams using leaf litter bags in the Dry Fork,
Gandy Creek, and Glady Fork watersheds, West Virginia, 2004………………….. 144
Table 7. Linear regression models of salamander diversity from leaf litter bag
sampling in streams in the Dry Fork, Gandy Creek, and Glady Fork watersheds,
West Virginia, 2004. Model rankings were based on Akaike’s Information
Criterion corrected for small sample size………………………………………….. 145
Chapter III: List of Figures
Figure 1. The 11-digit hydrologic units of lower Dry Fork, Gandy Creek, and
Glady Fork in the 8-digit Cheat River hydrologic unit located in West Virginia.
Numbered sites are streams with roads and alphanumeric sites are reference
streams……………………………………………………………………………... 146
Figure 2. Diagram of typical sampling point along a transect. Quadrats (1x1 m)
were searched on the (A) bank, in the (B) flow, and (C) dry channel (if present)… 147
Figure 3. Diagram of 2 transects sampled on each stream. Each transect had 9
sampling points and was either separated by a culvert or a 30 m stream segment… 148
ix
Chapter IV: List of Figures
Figure 1. The Dry Fork, Glady Fork, and Shavers Fork 10 digit hydrologic code
watersheds located in the Cheat River 8 digit watershed in eastern West Virginia.. 165
List of Appendices
Appendix 1. Results of filter classification of barrier types for stream salamander
passage through 120 culverts in the Dry Fork and lower Shavers Fork watersheds,
West Virginia, 2003………………………………………………………………... 167
Appendix 2. Locations of fragmented stream segments in the Shavers Fork and
Dry Fork watersheds, West Virginia, 2003……………………………………… 170
Appendix 3. Salamander captures at each site along transects in the Dry Fork,
Gandy Creek, and Glady Fork watersheds, West Virginia, 2004………………….. 176
Appendix 4. Salamander captures at each site using leaf litter bags in the Dry
Fork watershed, West Virginia, 2004……………………………………………… 177
Appendix 5. Logistic regression models with a negative binomial distribution
explaining the abundance of salamanders in the Dry Fork, Gandy Creek, and
Glady Fork watersheds, West Virginia, 2003……………………………………… 178
Appendix 6. Parameter estimates for liner regression models in the Dry Fork,
Gandy Creek, Glady Fork watersheds, West Virginia, 2004………………………. 184
Appendix 7. Parameter estimates for logistic models with a negative binomial
distribution on a reach scale in the Dry Fork, Gandy Creek, Glady Fork
watersheds, West Virginia, 2004………………………………………………… 189
Appendix 8. Parameter estimates for logistic models with a negative binomial
distribution on a stream scale in the Dry Fork, Gandy Creek, Glady Fork
watersheds, West Virginia, 2004…………………………………………………... 194
x
1
Chapter I:
Overview of Effects of Roads and Culverts on Salamanders and Descriptions
of Study Area
Ryan L. Ward
James T. Anderson
Division of Forestry
West Virginia University
P.O. Box 6125
Morgantown, WV 26505
2
Overview of effects of roads and culverts on salamanders and descriptions of study area
Abstract
Roads are a permanent and necessary part of the landscape and have many ecological
effects. Salamanders make up a large part of the faunal community in Appalachian forests.
Culverts are the most common type of crossing used when a road must cross a stream. Roads
and culverts can affect stream salamanders through alteration of habitat, population isolation, and
altering trophic levels. During the road planning process, thought should be given to prevent
disturbance to salamander communities and ecological processes. Minimizing impacts to
adjacent vegetation will prevent disruptions in microhabitat that have detrimental effects on
salamander communities. Minimizing sediment sources from roads is important for maintaining
interstitial spaces on which salamanders and other stream fauna depend. Providing the ability for
the passage of stream fauna will help maintain ecological integrity of systems and benefit
salamander communities by preventing the isolation of populations and allowing predators and
prey to coexist. Incorporating wildlife and ecology into future construction and road
maintenance is important to ensure coexistence of humans and wildlife.
Introduction and Justification
Roadways are a necessary component of human lives and a prominent feature on the
landscape. The need for roads is not likely to change, and therefore as wildlife managers and
environmental stewards we should strive to minimize their impacts on wildlife and their
ecosystems. Roads can have wide ranging ecological effects on the landscape. Forman and
Deblinger (2000) estimated an average width of 600 m for the zone of ecological impacts for a
busy 4-lane highway in Massachusetts. Forman (2000) extrapolated the ecological impacts of
the highway to determine that 1/5 of the land area in the United States was ecologically affected
This chapter written in the style of The Proceedings of the West Virginia Academy of Sciences.
3
by public roads. Angermeier et al. (2004) proposed examining the impacts of roads in three
phases: road construction, road presence, and urbanization. Road construction is often
considered when assessing environmental impacts, but road presence is often not considered and
urbanization is typically ignored (Angermeier et al. 2004).
Salamanders play important ecological roles in the faunal communities of Appalachian
stream and riparian ecosystems. Burton and Likens (1975a) conducted surveys of stream
salamanders at Hubbard Brook Experimental Forest in New Hampshire. Northern dusky
salamanders (Desmognathus fuscus) were found to occur at a density of 0.038 per m2 in 1970
and 0.031 in 1971. Northern two-lined salamander (Eurycea bislineata) adults were found at a
density of 0.037 per m2 and 0.022 per m2 and larva was found at a density of 0.498 per m2 and
0.756 per m2. Stream salamanders made up 6.5% of the total biomass of the salamander
community in the forest. The total biomass of salamanders was approximately twice the total
biomass of birds during the peak of the breeding season and approximately the same as the total
biomass of mice and shrews. Spight (1967) examined a collection of northern dusky
salamanders from a population in Gaston County in North Carolina. He found that the density
ranged from 0.43-1.42 individuals per m2 of streambed and the annual amount of biomass
produced ranged from 0.097-0.32 g per m2. The stream also contained 4 other species of
salamanders, which would increase the estimate of total biomass production of stream
salamanders. Burton and Likens (1975b) found annual energy flow through the salamander
population in the Hubbard Brook ecosystem to be 11,000 kcal/ha. This amount was roughly
equal to 20% of the energy flow through bird and mammal populations. Burton and Likens
(1975b) also determined that salamanders convert 60% of the ingested energy into new tissue,
4
resulting in salamander tissue containing higher levels of protein than bird and mammal tissue,
and making salamanders a good source of energy for predators.
Riparian areas provide important amphibian habitat and have distinct assemblages of
salamanders (Anderson et al. 2004). In Preston County, West Virginia, herpetofaunal
communities were found to differ in species distribution, size of individuals and biomass in
riparian versus upland areas (Spurgeon 2002). MacCulloch and Bider (1975) found that 75% of
the northern two-lined salamanders in a population in Quebec that survived the summer moved
less than 100 m from the stream. In western Oregon, upland and riparian sites had less than 40%
similarity between amphibian communities, and the total number of amphibians captured
declined as distance from the stream increased (McComb et al. 1993). Northern dusky
salamanders in Ohio used wet leaves, logs, and bark for microhabitat 24% of the time (Ashton
1975). In riparian forests on the Allegheny Plateau, mountain dusky salamanders
(Desmognathus ochrophaeus) and redback salamanders (Plethodon cineres) were frequently
found under rocks and downed wood (Moore et al. 2001). Increasing numbers of herpetofauna
that use riparian forests are being listed as endangered, threatened, or sensitive according to
federal, state, or agency mandates (Pauley et al. 2000).
Amphibians have been considered good indicators of biological stress in some situations
(Blaustein 1994). For instance, increased acidity in habitats has been shown to cause lethal and
sub-lethal effects on certain species (Dunson et al. 1992). Most researchers agree that further
long-term studies are needed to be able to separate natural fluctuations in amphibian populations
from human-induced changes (Pechmann et al. 1991, Dunson et al. 1992, Blaustein 1994).
Angermeier et al. (2004) stressed the need for post-construction impacts of roads and biotic
interactions of individuals, populations, and communities. More studies on environmental
5
impacts of roads will assist managers in assessing impacts for future construction projects
(Angermeier et al. 2004). Stream salamanders have been used to develop indices for stream
classification systems based on flow and ecological health (Ohio Environmental Protection
Agency 2002, Rocco et al. 2004).
The objectives of this chapter are to describe the area of our field study in detail and to
use existing literature to determine the effects of roads on streams, salamanders, salamander
habitat, and to explore other ways that roads may affect salamander communities such as altering
trophic levels present in streams.
Study Area
Culvert surveys were conducted in lower Shavers Fork of Cheat River watershed and Dry
Fork of Cheat River watershed (Figure 1). The watersheds were located in Randolph and Tucker
County in the eastern portion of West Virginia. Salamander surveys were conducted in the Dry
Fork watershed. Both areas were located in the Allegheny Plateau region. Green and Pauley
(1987) listed 9 different stream and streamside salamanders that may inhabit small streams in the
area (Table 1).
Climate
The average winter temperature in Randolph County is –0.6°C with an average daily
minimum of –6.7°C (Pyle et al. 1982). The average summer temperature is 19.4°C with an
average daily maximum of 26.7°C (Pyle et al. 1982). The average temperature for the year is
9.7°C with a daily average minimum of 3.1°C and daily average maximum of 16.4°C (Pyle et al.
1982). The average winter temperature in Parsons, West Virginia located in Tucker County is
0.3°C with an average daily minimum of –5.6°C (Losche and Beverage 1967). The average
summer temperature is 20.8°C with an average daily maximum of 28.2°C (Losche and Beverage
6
1967). The average temperature for the year is 10.8°C with a daily average minimum of 3.9°C
and daily average maximum of 17.8°C (Losche and Beverage 1967).
Randolph County averages 107.5 cm of rainfall each year and 150 cm of snowfall (Pyle
et al. 1982). The majority of the rainfall (55%) occurs between April and September (Pyle et al.
1982). Parsons, West Virginia, in Tucker County averages 125 cm of rainfall each year (Losche
and Beverage 1967). Snow cover is present an average of 37 days each year with an average
depth of 7.62 cm (Losche and Beverage 1967). Prevailing winds in Randolph County come
from the northwest (Pyle et al. 1982). Prevailing winds in Tucker County are westerly (Losche
and Beverage 1967).
Shavers Fork
The Shavers Fork watershed (Figure 2) consisted of 55,635 ha (Natural Resource
Conservation Service 2004). Shavers Fork flowed to the town of Parsons, WV where it
combined with Black Fork to form the Cheat River. Towns located in the Shavers Fork
watershed included Parsons, Bowden, Faulkner, and Bemis. Few large tributaries flowed into
Shavers Fork. Most tributaries were small first and second order streams that had a high
gradient. Elevations within the watershed ranged from 518 m to 1,472 m (West Virginia
Geographic Information Systems Technical Center 1999). The most abundant geologic map
units were the Pottsville group (24,882 ha), Mauch Chunk group (14,787 ha), and Chemung
group (11,448) (Cardwell et al. 1968). Major soil associations occurring were the DekalbBuchanan association, Calvin-high base substratum-Belmont-Meckesville association, DekalbCalvin-Belmont association, Gilpin association, Barbour-Pope-Sequatchie association, and the
Calvin association (Losche and Beverage 1967, Pyle et al. 1982). There were 38,543 ha (69.3%)
of forested land (Figure 3) in the watershed (West Virginia University Natural Resource
7
Analysis Center et al. 2002). Culvert surveys were conducted in the lower part of the watershed,
which had an extensive road network (Figure 4). The river maintained a cold and cool water
fishery.
Dry Fork
The Dry Fork watershed consisted of a total of 93,403 ha (Natural Resource Conservation
Service 2004). The river flowed near the town of Parsons, WV where it combined with
Blackwater River to form Black Fork. Towns and cities located in the Dry Fork watershed
included Hambleton, Hendricks, Gladwin, Red Creek, Alpena, Wymer, Harmon, Job, Whitmer,
and Laneville. Major tributaries to Dry Fork included Gandy Creek, Laurel Fork, Glady Fork,
and Red Creek. Minor tributaries to Dry Fork and its main tributaries were characterized by high
gradient mountain streams (Figure 5). The lower section of Dry Fork also constituted a
significant portion of the total watershed. Dry Fork supported a cool water fishery in some
sections and many of its tributaries supported cold and cool water fisheries. Culvert surveys and
salamander surveys were conducted in the watershed.
Gandy Creek and upper Dry Fork
Gandy Creek and the upper part of Dry Fork (Figure 6) added a total of 24,694 ha
(26.4%) to the Dry Fork watershed (Natural Resource Conservation Service 2004). Elevations
ranged from 612 m to 1,171 m (West Virginia Geographic Information Systems Technical
Center 1999). The most abundant geologic map units were the Mauch Chunk group (9,618 ha),
Greenbrier group (3,870 ha), and the Hampshire formation (7,206 ha) (Cardwell et al. 1968).
Soil associations included the Calvin-high base substratum-Belmont-Meckesville association and
the Dekalb-Berks-Calvin association (Pyle et al. 1982). Forested land (Figure 7) covered 19,420
ha (78.6%) of the Gandy Creek and upper Dry Fork drainage area (West Virginia University
8
Natural Resource Analysis Center et al. 2002). Gandy Creek flowed into a cave called the Sinks
of Gandy in the upper part of its drainage area. It reemerged a short distance downstream. Parts
of Gandy Creek supported an important cold-water fishery. Culvert surveys (Figure 8) and
salamander surveys (Figure 9) were conducted on small tributaries to Gandy Creek.
Laurel Fork
Laurel Fork (Figure 10) had a drainage area of 15,613 ha (16.7%) of the Dry Fork
watershed (Natural Resource Conservation Service 2004). Elevations ranged from 701 m to
1,319 m (West Virginia Geographic Information Systems Technical Center 1999). The most
abundant geologic map units were the Hampshire formation (9,984 ha), Mauch Chunk group
(1,680 ha), and Pocono group (1,527 ha) (Cardwell et al. 1968). Soil associations included the
Dekalb-Berks-Calvin association, Dekalb-Gilpin association, and the Calvin association (Losche
and Beverage 1967, Pyle et al. 1982). Forested land (Figure 11) covered 12,654 ha (81.0%) of
the Laurel Fork drainage area (West Virginia University Natural Resource Analysis Center et al.
2002). The upper part of this watershed included the Laurel Fork Wilderness Area. The
tributary supported a cold and cool water fishery. Due to a lack of state roads, Laurel Creek was
excluded from culvert and salamander surveys.
Glady Fork
Glady Fork (Figure 12) drained 16,439 ha (17.6%) of the Dry Fork watershed (Natural
Resource Conservation Service 2004). Elevations ranged from 640 m to 1,390 m (West Virginia
Geographic Information Systems Technical Center 1999). The Hampshire formation (7,185 ha),
Chemung group (4,719 ha), and Mauch Chunk group (2,289 ha) were the most abundant
geologic map units (Cardwell et al. 1968). Soil associations included the Calvin-high base
substratum-Belmont-Meckesville association, Dekalb-Berks-Calvin association, Calvin
9
association, Barbour-Pope-Sequatchie association, Dekalb-Gilpin association, and DekalbCalvin-Belmont association (Losche and Beverage 1967, Pyle et al. 1982). Forested land (Figure
13) covered 15,157 ha (92.2%) of the Glady Fork drainage area (West Virginia University
Natural Resource Analysis Center et al. 2002). The tributary supported both cold and cool water
fisheries. Culvert surveys were conducted on some sites in the tributary’s watershed to aid a
concurrent study. Salamander sampling was conducted on a culverted tributary and on a
nonculverted reference stream draining into Glady Fork (Figure 14).
Red Creek
Red Creek (Figure 15) contributed 15,890 ha (17.0%) to the Dry Fork watershed (Natural
Resource Conservation Service 2004). Elevations ranged from 640 m to 1,440 m (West Virginia
Geographic Information Systems Technical Center 1999). The most abundant geologic map
units were the Conemaugh group (4,251 ha), Pottsville group (3,672 ha), and Allegheny
formation (3,372 ha) (Cardwell et al. 1968). Soil associations included the very stony landErnest-Brinkerton-Leetonia association, very stony land-Dekalb association, and the DekalbCalvin-Belmont association (Losche and Beverage 1967). Forested land (Figure 16) covered
13,710 (86.3%) of the Red Creek drainage area (West Virginia University Natural Resource
Analysis Center et al. 2002). Red Creek drained the Dolly Sods Wilderness area. The main stem
of Red Creek was listed as impaired by the West Virginia Department of Environmental
Protection according to the Clean Water Act Section 303d (West Virginia Department of
Environmental Protection 2003). It was excluded from culvert and salamander surveys due to
poor water quality.
10
Lower Dry Fork
The lower part of Dry Fork (Figure 17) consisted of 20,766 ha (22.2%) of the total Dry
Fork watershed (Natural Resource Conservation Service 2004). Elevations in this section ranged
from 518 m to 1,250 m (West Virginia Geographic Information Systems Technical Center 1999).
The Pottsville group (8,421 ha), Mauch Chunk group (3,501 ha), and Hampshire formation
(2,908 ha) were the most common geologic map units (Cardwell et al. 1968). Soil associations
included the Dekalb-Calvin-Belmont association and the Calvin association (Losche and
Beverage 1967). Forested land (Figure 18) covered 18,896 ha (91.0%) of this section of the Dry
Fork watershed (West Virginia University Natural Resource Analysis Center et al. 2002). Otter
Creek was a tributary in this section that drained the Otter Creek Wilderness Area. Lower Dry
Fork supported a cool water fishery and some larger tributaries supported cold water species.
Culvert surveys (Figure 19) and salamander surveys (Figure 20) were both conducted in this
portion of the Dry Fork watershed.
Literature review
Roads and culverts have direct and indirect effects on salamanders through mortality,
creation of movement barriers, and alteration of habitat. Effects of roads on other members of
the faunal community such as fish also can have impacts on salamanders.
Salamanders and habitat
Mortality from roads significantly affects amphibian populations. Fahrig et al. (1995)
found that the numbers of road killed frogs and toads increased with traffic density while
population densities of frogs and toads decreased with traffic density. They concluded that the
decrease in density was due to the increased mortality from the road traffic.
11
Roads also can affect the shading of streams (Figure 21). Miller et al. (1997) found a
reduced basal area of overstory trees in the immediate vicinity of road crossings in
Pennsylvania. Plants in road clearings are often disturbed with mowing, which prevents
succession and keeps the area at a lower seral stage (Mader 1984). Canopy removal can have a
negative effect on salamanders. In Pennsylvania, larval two-lined and adult northern dusky
salamander abundance was positively correlated with higher amounts of percent cover (Bast and
Maret 1998). Most salamander research has been conducted on terrestrial salamanders. Mitchell
et al. (1997) found that amphibians in central Appalachian forests were more abundant in areas
with mature hardwoods compared to recent clearcuts. In Washington, most salamanders
declined when forests were clearcut (Grialou et al. 2000). Petranka et al. (1994) found that
terrestrial salamanders in North Carolina were eliminated or greatly reduced when forests were
clearcut. Pough et al. (1987) found that the abundance of salamanders in New York was much
greater in old growth forests compared to recently disturbed sites. Pough et al. (1987) also found
that the density of understory vegetation and depth of leaf litter was positively correlated with
surface activity of salamanders. In Pennsylvania, Ross et al. (2000) concluded that salamander
abundance increased with increasing tree basal area. In Redwood National Park, California,
Bury (1983) found that logging and the removal of forest canopy had long-term effects of the
herpetofaunal communities, benefiting a few species while being detrimental to most of the
forest-dependent species. Bury (1983) found that old growth sites had more individuals, greater
biomass, and different species composition than logged sites. Road edges are long-term
permanent edge, and they are disturbed frequently compared to other types of edges such as clear
cuts (Reed et al. 1996). In the Medicine Bow-Routt National Forest of southwestern Wyoming,
roads created 1.54-1.98 times as much edge habitat as clear cuts (Reed et al. 1996). New edges
12
have been shown to cause shifts in the microclimates, which alter native plant and animal
communities (Collinge 1996). Changes in microclimate are important to salamanders.
Plethodontids rely on cutaneous respiration, which restricts their activities to areas with moist
microclimates, and make them susceptible to the loss of forest canopy (DeMaynadier and Hunter
1999, Feder 1983). In Maine, edge effects on woodland amphibian populations extended for 2535 m into the forest (DeMaynadier and Hunter 1998). Breaks in the vegetation also limit the
movement of salamanders. DeMaynadier and Hunter (1999) found that juvenile wood frogs
(Rana sylvatica) and spotted salamanders (Ambystoma maculatum) used closed canopy forests
for emigration and dispersal. The gap in continuous vegetation caused by roads and road rightof-ways is likely to have a negative impact on salamander populations.
Roads have been shown to increase sediments in streams. Burns (1972) showed that
logging and road construction increased the amount of sediment in some California streams. In
Pennsylvania, streams below road crossings had higher levels of fine sediments in the streambed
(Miller et al. 1997). Beschta (1978) found that roads were a major source of sediment
production in logged watersheds. Traffic intensity also plays an important factor in sediment
production. Reid and Dunne (1984) found that heavily used roads contributed 130 times more
sediment than abandoned roads. Road surface also has an effect. Reid and Dunne (1984) found
that paved roads yielded less than 1% of the sediment yield of gravel road under heavy use.
Roads with under-sized culverts (Figure 22) can cause a backup of water, which will cause the
stream to deposit more sediment upstream of the road in addition to sediment from the road
deposited downstream (Forman et al. 2003). Negative effects of sediment are greatest in low
gradient streams where it is more likely to accumulate (Corn and Bury 1989, Murphy and Hall
13
1981). In New Hampshire, Lowe and Bolger (2002) found negative associations between
stream embeddedness and salamanders.
Channelization of the stream often occurs up and downstream from road crossings
(Figure 23). Forman and Deblinger (2000) found that nearly all stream crossings along a fourlane road in Massachusetts showed evidence of channelization. Two-thirds of the streams had
evidence of channelization in the upstream portion, and one-third of the streams had evidence in
the downstream portion (Forman and Deblinger 2000). The negative effects of channelization
include destruction of riparian vegetation, water temperature increases due to lack of shade and
increased surface area, a lack of pool and riffle habitat, no protection for aquatic organisms at
bankfull velocities, and increased likelihood of bank erosion (Beschta and Platts 1986). Stream
habitat degradation was apparent in Idaho where 38% of 45 study streams had been altered by
some form of channelization, with unaltered streams producing 1.4 to 112 times as much
biomass of game fish as altered streams (Irizarry 1969).
Habitat Fragmentation
Roads serve as animal barriers because they create breaks in the microclimate, create
disturbance, have environmentally unstable verges, and result in the death of individuals through
direct mortality (Mader 1984). Roads create breaks in the continuous landscape and isolate
populations of animals. When small populations become isolated they become more vulnerable
to extinction from factors such as inbreeding depression, demographic events, and environmental
events (Mills and Smouse 1994). In small populations, demographic stochasticity becomes a
more important factor in the persistence of populations compared to large populations (Lande
1993). Gibbs (1998a) found that amphibians with low population densities were the most
sensitive to fragmentation. Gibbs (1998a) also stated that amphibians considered “high
14
dispersers” were the least resistant to habitat fragmentation. High dispersers possibly end
migration in unsuitable areas or become stranded in open areas resulting in failed recruitment
into breeding populations. As the size of a habitat fragment decreases so does the species
richness of the fragment (Collinge 1996). Gibbs (1998b) found populations of redback
salamanders had measurable differences in genetic diversity compared to populations found in a
continuous landscape. Good movement corridors allow for genetic interchange, allow
populations to move in response to changes and disasters in their environment, and recolonize
habitats where previous populations had been extirpated (Beier and Loe 1992). Lowe and
Bolger (2002) found that population connectivity could help buffer populations of spring
salamanders (Gyrinophilus porphyriticus) from disturbance. Connectivity of streams for
salamanders is also important because many semi-aquatic adults will move downstream to
deposit eggs in hydrologically stable areas then move upstream to exploit productivity in
headwater streams (Jackson 2003).
Culverts and Salamander Passage
Culverts are used to control the flow path of road drainage and stream channels and keep
the water separate from the road (Adair et al. 2002). Culverts are a common type of stream
crossing used by transportation planners. Common types of culverts (Figure 24) include circular,
open-bottom arch, pipe arch, and box (Taylor and Love 2003). Common construction materials
used in culverts include corrugated steel pipe (Figure 25), structural steel pipe, aluminum,
plastic, concrete (Figure 26), and wood (Taylor and Love 2003).
Most studies on the passage of culverts have concentrated on fish. Fish movement
through culverts was found to be lower than at other types of crossings (Warren and Pardew
1998). Resident trout in streams in Montana were able to pass through highway culverts ranging
15
from 42 m to 93 m long with slopes ranging from 0.2% to 4.4% (Belford and Gould 1989). At
discharges from 0.0113 to 0.017 m3/s, juvenile salmonids ranging from 50 mm to 100 mm were
able to pass through a 90 cm diameter culvert that was 9 m long, had 10% slope, and contained
offset baffles (Bryant 1981). Culverts made of smooth pipe allowed coho salmon
(Oncorhynchus kisutch) fry and fingerlings to pass at flow velocities up to the fish’s swimming
ability, but turbulence in corrugated pipes prevented passage at velocities above 0.61 m/s
(Powers et al. 1997). Culverts have high flow velocities when compared to other types of
crossings, and Warren and Pardew (1998) found velocity was inversely related to fish
movement. Thompson and Rahel (1998) found that brook trout (Salvelinus fontinalis) were
prevented from moving upstream past culverts with drops of 0.50 m to 0.75 m.
There is a lack of studies on the aspects of passage related to salamanders. D’Aout and
Aerts (1997) found that adult axolotls (Ambystoma mexicanum) were less efficient swimmers
than most fishes. Fitzpatrick et al. (2003) found a maximum velocity of 0.295 m/s for tiger
salamander larvae (Ambystoma tigrinum mavortium). Maximum burst swimming speeds of adult
red salamanders (Pseudotriton ruber) were approximately 0.4 m/s (Marvin 2003). Maximum
velocities for tiger salamander larvae are lower than adult fish of similar length (Fitzpatrick et al.
2003). Poor swimming performance relative to fish would suggest that culvert velocities would
have a greater impact on salamanders, but terrestrial capabilities of salamanders could possibly
allow them to navigate through or around culverts. Negative effects of culverts have been
observed on populations of Pacific giant salamanders (Dicamptodon tenebrosus) (Richardson
2002).
Common conditions that block fish passage through culverts include velocities that are
too high (Figure 27), lack of depth of water inside (Figure 28), lack of resting pool below culvert
16
(Figure 29), and outlet drops that are beyond the jumping ability of fish (Figure 30) (Taylor and
Love 2003). Also the buoyant forces and weight of fish become problems for fish passage
through culverts with high slopes (Behlke 1987). Some solutions to passage problems do exist.
Water velocities have been slowed with the addition of corrugations and baffles inside culverts,
which increase roughness (Taylor and Love 2003). The presence of streambed material inside
culverts (Figure 31) causes variability in velocities that can allow fish passage (Kahler and Quinn
1998). Bottomless culverts simulate natural stream conditions and promote aquatic organism
and fish passage (Adair et al. 2002). Weirs can be used to adjust stream gradients at the inlet and
outlets of culverts, compensating for large drops and hydraulic forces exceeding the physical
capabilities of fish (Taylor and Love 2003). Embedding the culvert also allows for a stream
channel to form inside the culvert, mimicking the rest of the stream allowing sufficient depth of
water (Taylor and Love 2003). The bed material in a stream may be the most biologically
significant characteristic of the stream (Cummins 1974). Fish use bed material for spawning,
cover, and foraging (Beschta and Platts 1986). Salamanders use the channel substrate for refuge
and foraging (Moore et al. 2001). Road networks affect peak flows and may trigger or stop
debris flows that determine the bed material of a stream (Jones et al. 2000). White (2004) found
that undersized culverts occurred at 75% of the crossings in the upper Cheat River basin, leading
to sedimentation, blockage, and conveyance problems.
Fish interactions
Fish are often the top predators in aquatic systems (Figure 32). The effects of culverts on
fish may cause a possible secondary effect on salamanders. If culverts prevent fish movement
into upstream stream reaches it may alter trophic levels and allow some stream salamanders to
increase in the absence or reduction of predation by fish. In these streams Plethodontid
17
salamanders can flourish and assume the role of top predator. Interactions between predators
and their prey have been well studied, however in the case of fish and salamanders a predatorprey relationship exists along with a competitive predator relationship. Salamanders suffer from
direct predation by fish along with predator aggression, interference competition, and
exploitative competition. This competitive relationship is asymmetrical, highly in favor of
predatory fish (Resetarits 1995).
Exploitative Competition
Many studies list competition for food as a possible mechanism that reduces the survival
or growth rate of salamanders in the presence of fish (Resetarits 1991, Barr and Babbitt 2002).
Smith et al. (1999) found the abundance of red-spotted newts (Notophthalmus virirdescens)
decreased in the presence of bluegill sunfish (Lepomis macrochirus). Since the red-spotted newt
is unpalatable to fish, they concluded that the decrease in abundance was because of exploitative
competition. Newt densities increased in fishless areas, which also contained more abundant
food in the form of Daphnia spp. Species in other studies followed a similar pattern of habitat
shift but the avoidance of predators was often implicated as the main reason (Resetarits 1991,
Barr and Babbitt 2002).
Demographic Effects
Fish can alter the demographic rates of salamanders. The survival rate of salamander
populations is the most obvious rate altered by fish predation. Lowe and Bolger (2002) surveyed
headwater streams in New Hampshire and found that the presence of brook trout caused a
decline in abundance of northern spring salamanders and speculated that direct predation was the
cause. Resetarits (1991) created experimental streams to examine the effects of brook trout on
northern spring and two-lined salamanders. Brook trout caused a 35% drop in the survival rate
18
of larval spring salamanders, but had no significant effect on larval two-lined salamander
survival. The author attributed the reduction in the survival rate of spring salamander larvae to
territorial behavior of brook trout and attempted predation by brook trout. Reserarits (1995)
found that the presence of fingerling brook trout reduced the survival rate of northern spring
salamander larvae by 50%, and attributed this reduction to the strong effects of aggression and
interference competition between the two vertebrate species.
The reproductive rates of salamanders are closely tied to body size (Semlitsch et al.
1988), and the growth rate of larval salamanders can be affected by predation risk (Ziemba et al.
2000). This can affect the overall fitness of an individual. Larvae of the pond-breeding, mole
salamander (Ambystoma talpoideum) delayed sexual maturity until reaching a larger size when in
the presence of bluegill sunfish (Jackson and Semlitsch 1993). Fish presence also significantly
reduces the growth rate of larval spotted salamanders (Figiel and Semlitsch 1990). Tyler et al.
(1998) created experimental ponds with rainbow trout (Oncorhynchus mykiss) to examine
growth rates of the larvae of northwestern salamanders (Ambystoma gracile) and long-toed
salamanders (Ambystoma macrodactylum). The authors found that the presence of trout
significantly lowered the snout-vent lengths (SVL) of larvae by the end of the experiment.
Moore et al. (1996) showed that chemical cues from green sunfish (Lepomis cyanellus) delayed
hatching of eggs from the streamside salamander (Ambystoma barbouri). Barr and Babbitt
(2002) found that trout occupied streams generally had larger two-lined salamander larvae
compared to streams without trout. They believed that small larvae suffered from high predation
rates or possibly cannibalism from larger larvae in more restricted spaces used because of
predator avoidance. Resetarits (1995) saw a 35% decrease in growth of spring salamanders in
artificial streams with brook trout. In the same experiment, surviving larval two-lined
19
salamanders had a lower mean size than larval in artificial stream lacking trout. This reduction
in size was possibly caused by decreased foraging behavior to avoid predation. In this
experiment, activity by two-lined salamander larvae decreased 36% in the presence of brook
trout. Resetarits (1995) saw a reduction in growth greater than 90% for mass and 44% for SVL
of larval northern spring salamanders.
Lowe and Bolger (2002) found that streams with a confluence of a first-order stream had
higher abundances of spring salamander compared to isolated streams. The authors suggested
that population connectivity may help offset stream-scale disturbances that negatively affect
spring salamanders including predation by brook trout. Resetarits (1995) suggested that a
source-sink relationship might exist between upstream (trout-free) sections of streams and
downstream (trout-inhabited) sections when exploring mechanisms for the persistence of spring
salamander populations in the presence of brook trout. Pilliod (2001) speculated that high
mountain lakes in Idaho probably once supported large amphibian populations, but now serve as
sinks because of introduced trout. The lakes are supported with immigrants from nearby fishless
sites. Amezcua and Holyoak (2000) examined the predator-prey relationships of a protist
predator and prey. The experiment was conducted in a laboratory in a microcosm. The
microcosm was divided in half and prey organisms were able to disperse freely across the
divider. These individuals performed a rescue effect of the area of the microcosm subject to
predation. Populations with the divided population of prey persisted three times as long as
populations of prey in undivided microcosms. This experiment points to the importance of
immigration in helping prey populations persist. If culverts prevent fish movement, streams with
unnatural, thriving salamander populations may be created. These streams may serve as sources
and possibly alter genetic diversity of total salamander populations.
20
Spatial Distribution
Fish affect the distribution and habitat use of salamanders. Low-order streams are patchy
habitats that can be divided many different ways including geomorphic channel units, open or
closed canopy, and variation in water depths. Predators are often distributed in relation to prey
density and quality (Hildrew and Townsend 1982, Holt 1987). For salamanders, the threat of
predation can force them into marginal habitats from the perspective of prey abundance. This
response has been seen in fish where patches of lower food abundance were selected to avoid
predation risk (Holbrook and Schmitt 1988). Smith et al. (1999) found that bluegill sunfish
caused red-spotted newts to shift their use to areas lacking fish that had higher densities of food.
Barr and Babbitt (2002) sampled streams in the White Mountain National Forest in New
Hampshire for northern two-lined salamanders and brook trout. They found that two-lined
salamander larvae were less abundant in sections of stream with brook trout. In the presence of
brook trout two-lined salamander larvae was found at higher densities in areas of the stream with
high boulder cover and low amounts of sand and bare rock (Barr and Babbitt 2002). In an
enclosure experiment conducted by Barr and Babbitt (2002) two-lined salamander larvae had a
higher survival rate when cobbles were available for cover instead of just gravel, pebbles, and
sand. Long-toed salamander and northwestern salamander larvae in ponds with fish favored rock
substrates and they did not use open water and vegetative cover like larvae in ponds without fish
(Tyler et al. 1998). Habitat complexity has been shown to decrease fish predation efficiency
(Crowder and Cooper 1982). Kats and Sih (1992) found that streamside salamanders avoid fish
during oviposition to reduce the mortality of eggs and larvae. They deposit higher densities of
eggs in pools without fish compared to pools with fish. Resetarits (1991) found that spring
salamander and two-lined salamander larvae restricted their habitat use to shallow areas of the
21
experimental streams in the presence of brook trout. Resetarits (1995) showed that spring
salamanders switch from a uniform distribution to more shallow water habitats in the presence of
brook trout fingerlings.
Sih and Wooster (1994) stated that if predators promote prey emigration from patches, a
large positive effect will result and densities in patches with predators will decrease. If prey
emigration is suppressed from patches, prey immigration may outweigh both predation and
emigration and result in a negative predator effect in the form of increased prey densities in the
presence of predators (Sih and Wooster 1994). Fish appear to have a positive effect on
salamanders, causing emigration rates to increase and densities to decrease in the presence of
predators, resulting in areas that contain fish having low salamander densities.
Predatory fish in a river can cause the fragmentation of prey populations in small
tributaries lacking predators (Fraser et al. 1995). Similar processes may happen to salamanders
where areas inaccessible to predatory fish are the major habitats used by salamanders. These
habitats may include small tributaries, and shallow areas that fish do not prefer due to lack of
food or the predation risk for fish by avian or other predators outside the stream system.
Conclusions
Road networks can have significant effects on salamander communities. Population
isolation, habitat alteration, and trophic level alteration may be the most important effects.
Angermeier et al. (2004) hypothesized that road density is correlated with an increase in
predominance in species tolerant to silt, metals, petroleum products and salt, and species that are
good colonizers.
Road crossings of streams should be designed to impacts stream functions and
communities as little as possible. Incorporating geomorphic processes into culvert design should
22
improve crossings (White 2004). Transportation planning should also include measures to
minimize ecological impacts of roads. Stream salamander communities would benefit from road
crossings that allowed for population connectivity, maintained forest canopy, prevented
sedimentation of streams, and allowed for ecological processes to occur.
Literature Cited
Adair, S, M. L. Dereske, J. Doyle, A. Edwards, S. Jacobson, R. Jemison, L. Lewis, W. Melgin.
C. Napper, T. Ratcliff, and T. Warhol. 2002. Management and techniques for riparian
restorations. United States Department of Agriculture., Forest Service, Rocky Mountain
Research Center. Research Note. No. RMRS-CTR-102.
Amezcua, A. B. and M. Holyoak. 2000. Empirical evidence for predator-prey source-sink
dynamics. Ecology 81:3087-3098.
Anderson, J. T., J. D. Osbourne, and R. L. Ward. 2004. Integrating riparian restoration to
promote wildlife habitat with natural stream channel design on mine habitat with natural
stream channel design on mine land habitats. Proceedings of the Twenty-first Annual
American Society of Mining and Reclamation Meeting 21:47-73.
Angermeier, P. L., A. P. Wheeler, and A. E. Rosenberger. 2004. A conceptual framework for
assessing impacts of roads on aquatic biota. Fisheries 29:19-29.
Ashton, R. E. 1975. A study of movement, home range, and winter behavior of Desmognathus
fuscus (Rafinesque). Journal of Herpetology 9:85-91.
Barr, G. E. and K. J. Babbitt. 2002. Effects of biotic and abiotic factors on the distribution and
abundance of larval two-lined salamanders (Eurycea bislineata) across spatial scales.
Oecologia 133:176-185.
23
Bast, K. and T. J. Maret. 1998. Ecological correlates of stream salamander abundance and
distribution. Journal of the Pennsylvania Academy of Science 71:164 (abstract).
Behlke, C. E. 1987. Hydraulic relationships between swimming fish and water flowing in
culverts. in Proceedings of the 2nd International Conference on Cold Region
Environmental Engineering, CSCE-ASCE, University of Alberta, Edmonton, Alberta. pp.
112-132.
Beier, P. and S. Loe. 1992. A checklist for evaluating impacts to wildlife movement corridors.
Wildlife Society Bulletin 20:434-440.
Belford, D. A. and W. R. Gould. 1989. An evaluation of trout passage through six highway
culverts in Montana. North American Journal of Fisheries Management 9:437-445.
Beschta, R. L., and W. S. Platts. 1986. Morphological features of small streams: significance
and function. Water Resources Bulletin 22:369-379.
Beschta, R. L. 1978. Long-term patterns of sediment production following road construction
and logging in the Oregon Coast Range. Water Resources Research 14:1011-1016.
Blaustein, A. R. 1994. Chicken Little of Nero’s fiddle? A perspective on declining amphibian
populations. Herpetologica 50:85-97.
Bryant, M. D. 1981. Evaluation of a small diameter baffled culvert for passing juvenile
salmonids. U. S. Department of Agriculture, Forest Service, Pacific Northwest Forest
and Range Experiment Station. Research Note. No. PNW-384.
Burns, J. W. 1972. Some effects of logging and associated road construction on northern
California streams. Transactions of the American Fisheries Society 101:1-17.
Burton, T. M. and G. E. Likens. 1975a. Salamander populatons and biomass in the Hubbard
Brook Experimental Forest, New Hampshire. Copeia 1975:541-546.
24
Burton, T. M. and G. E. Likens. 1975b. Energy flow and nutrient cycling in salamander
populations in the Hubbard Brook Experimental Forest, New Hampshire. Ecology
56:1068-1080.
Bury, R. B. 1983. Differences in Amphibian populations in logged and old growth redwood
forest. Northwest Science 57:167-178.
Cardwell, D. H., R. B. Erwin, and H. P. Woodward. 1968. Geologic map of West Virginia.
West Virginia Geological and Economic Survey. 1:250,000 scale. West Virginia
Department of Environmental Protection and United States Geologic Survey Water
Resources Division. 1998 (digitized). UTM Zone 17 NAD 27.
Collinge, S. K. 1996. Ecological consequences of habitat fragmentation: implications for
landscape architecture and planing. Landscape and Urban Planning 36:59-77.
Corn, P. S. and R. B. Bury. 1989. Logging in western Oregon: responses of headwater habitats
and stream amphibians. Forest Ecology and Management 29:39-57.
Crowder, L. B. and W. E. Cooper. 1982. Habitat structural complexity and the interaction
between bluegills and their prey. Ecology 63:1802-1813.
Cummins, K.W. 1974. Structure and function of stream ecosystems. BioScience 24:631-641.
D’Aout, K. and P. Aerts. 1997. Kinematics and efficiency of steady swimming in adult axolotls
(Ambystoma mexicanum). Journal of Experimental Biology 200:1863-1871.
DeMaynadier, P. G. and M. L. Hunter, Jr. 1999. Forest canopy closure and juvenile emigration
by pool-breeding amphibians in Maine. Journal of Wildlife Management 63:441-450.
DeMaynadier, P. G. and M. L. Hunter, Jr. 1998. Effects of silvicultural edges on the
distribution and abundance of amphibians in Maine. Conservation Biology 12:340-352.
25
Dunson, W. A., R. L. Wyman, and E. S. Corbett. 1992. A symposium on amphibian declines
and habitat acidification. Journal of Herpetology 26:349-352.
Fahrig, L., J. H. Pedlar, S. E. Pope, P. D. Taylor, and J. F. Wegner. 1995. Effect of road traffic
on amphibian density. Biological Conservation 73:177-182.
Feder, M. E. 1983. Integrating the ecology and physiology of plethodontid salamanders.
Herpetologica 39:291-310.
Figiel, Jr., C. R. and R. D. Semlitsch. 1990. Population variation in survival and metamorphosis
of larval salamanders (Ambystoma maculatum) in the presence and absence of fish
predation. Copeia 1990:818-826.
Fitzpatrick, B. M., M. F. Benard, and J. A Fordyce. 2003. Morphology and escape performance
of tiger salamander larvae (Ambystoma tigrinum mavortium). Journal of Experimental
Zoology 297A:147-159.
Forman, R. T. 2000. Estimate of the area affected ecologically by the road system in the United
States. Conservation Biology 14:31-35.
Forman, R. T. and R. D. Deblinger. 2000. The ecological road-effect zone of a Massachusetts
(U.S.A.) suburban highway. Conservation Biology 14:36-46.
Forman, R. T., D. Sperling, J. A. Bissonette, A. P. Clevenger, C. D. Cutshall, V. H. Dale, L.
Fahrig, R. France, C. R. Goldman, K. Heanue, J. A. Jones, F. J. Swanson, T. Turrentine,
and T. C. Winter. 2003. Road Ecology Science and Solutions. Island Press, Washington
D.C., U.S.A.
Fraser D. F., J. F. Gilliam, and T. Yip-Hoi. 1995. Predation as an agent of population
fragmentation in a tropical watershed. Ecology 76:1461-1472.
26
Gibbs, J. P. 1998a. Distribution of woodland amphibians along a forest fragmentation gradient.
Landscape Ecology 13:263-268.
Gibbs, J. P. 1998b. Genetic structure of redback salamander Plethodon cinereus populations in
continuous and fragmented forests. Biological Conservation 86:77-81.
Green N. B. and T. K. Pauley. 1987. Amphibians and reptiles in West Virginia. University of
Pittsburgh Press, Pittsburgh, PA, USA.
Grialou, J. A., S. D. West, and R. N. Wilkins. 2000. The effects of forest clearcut harvesting
and thinning on terrestrial salamanders. Journal of Wildlife Management 64:105-113.
Hildrew, A. G. and C. R. Townsend. 1982. Predators and prey in a patchy environment: a
freshwater study. Journal of Animal Ecology 51:797-815.
Holbrook, S. J. and R. J. Schmitt. 1988. The combined effects of predation risk and food reward
on patch selection. Ecology 69:125-134.
Holt, R. D. 1987. Prey communities in patchy environments. Oikos 50: 276-290.
Irizarry, R . A. 1969. The effects of stream alteration in Idaho. Idaho Fish and Game
Department, Job Completion Report, Project F 55-R-2.
Jackson, M. E. and R. D. Semlitsch. 1993. Paedomorphosis in the salamander Ambystoma
talpoideum: effects of a fish predator. Ecology 74:342-350.
Jackson S. D. 2003. Ecological considerations in the design of river and stream crossings in C.
L. Irwin, P. Garrett, and K. P. McDermott, editors, 2003 Proceedings of the International
Conference on Ecology and Transportation. Raleigh, North Carolina: Center for
Transportation and the Environment, North Carolina State University.
27
Jones, J.A., F.J. Swanson, B.C. Wemple, and K.U. Snyder. 2000. Effects of roads on
hydrology, geomorphology, and disturbance patches in stream networks. Conservation
Biology 14:76-85.
Kahler, T. H. and T. P. Quinn. 1998. Juvenile and resident salmonid movement and passage
through culverts. Final Research Report, Research project T9903, Task 96. Washington
State Transportation Center.
Kats, L. B. and A. Sih. 1992. Oviposition site selection and avoidance of fish by streamside
salamanders (Ambystoma barbouri). Copeia 1992:468-473.
Lande, R. 1993. Risks of population extinction from demographic and environmental stochasity
and random catastrophes. The American Naturalist 142:911-927.
Losche, C. K., and W. W. Beverage. 1967. Soil survey of Tucker County and part of northern
Randolph County, West Virginia. United States Department of Agriculture, Soil
Conservation Service and Forest Service.
Lowe, W. H. and D. T. Bolger. 2002. Local and landscape-scale predictors of salamanders
abundance in New Hampshire headwater streams. Conservation Biology 16:183-193.
MacCulloch, R. D. and J. R. Bider. 1975. Phenology, migrations, circadian rhythm and the
effect of precipitation of the activity of Eurycea b. bislineata in Quebec. Herpetologica
31:433-439.
Mader, H. J. 1984. Animal habitat isolation by roads and agricultural fields. Biological
Conservation 29:81-96.
Marvin, G. A. 2003. Aquatic and terrestrial locomotor performance in a semiaquatic
plethodontid salamander (Pseudotriton ruber): influence of acute temperature, thermal
acclimation, and body size. Copeia 2003:704-713.
28
McComb, W. C., K, McGarigal, and R. G. Anthony. 1993. Small mammal and amphibian
abundance in streamside and upslope habitats of mature Douglas-fir stands, western
Oregon. Northwest Science 67:7-15.
Miller, Jr., R. L., D. R. Dewalle, R. P. Brooks, and J. C. Finley. 1997. Long-term impacts of
forest road crossings of wetlands in Pennsylvania. Northern Journal of Applied Forestry
14:109-116.
Mills, L. S. and P. E. Smouse. 1994. Demographic consequences of inbreeding in remnant
populations. The American Naturalist 144:412-431.
Mitchell, J. C., S. H. Rinehart, J. F. Pagels, K. A. Buhlmann, and C. A. Pague. 1997. Factors
influencing amphibian and small mammal assemblages in central Appalachian forests.
Forest Ecology and Management 96:65-76.
Moore, A. L., C. E. Williams, T. H. Martin, and W. J. Moriarity. 2001. Influence of season,
geomorphic surface, and cover item on capture, size, and weight, of Desmognathus
ochrophaeus and Plethodon cinereus in Allegheny Plateau riparian forests. The
American Midland Naturalist 145:39-45.
Moore, R. D., B. Newton, A. Sih. 1996. Delayed hatching as a response of streamside
salamander eggs to chemical cues from predatory sunfish. Oikos 77:331-335.
Murphy, M. L. and J. D. Hall. 1981. Varied effects of clear-cut logging on predators and their
habitat in small streams of the Cascade Mountains, Oregon. Canadian Journal of
Fisheries and Aquatic Science 38:137-145.
Natural Resource Conservation Service (NRCS). 2004. Watershed boundary data set draft 10
digit. UTM Zone 17 NAD 83.
29
Ohio Environmental Protection Agency. 2002. Technical report: Ohio’s primary headwater
streams-fish and amphibian assemblages. Ohio Environmental Protection Agency,
Division of Surface Water, Columbus, Ohio, USA.
Pauley, T. K., J. C. Mitchell, R. R. Buech, and J. J. Moriarty. 2000. Ecology and management
of riparian habitats for amphibians and reptiles in E. S. Verry, J. W. Hornbeck, and C. A.
Dolloff, editors, Riparian Management in Forests of the Continental Eastern United
States. Lewis Publishers, Boca Raton, Florida.
Pechmann, J. H. K., D. E. Scott, R. D. Semlitsch, J. P. Caldwell, L. J. Vitt, and J. W. Gibbons.
1991. Declining amphibian populations: the problem of separating human impacts from
natural fluctuations. Science 253:892-895.
Petranka, J. W., M. P. Brannon, M. E. Hopey, and C. K. Smith. 1994. Effects of timber
harvesting on low elevation populations of southern Appalachian salamanders. Forest
Ecology and Management 67:135-147.
Pilliod, D.S. 2001. Ecology and Conservation of high-elevation amphibian populations in
historically fishless watersheds with introduced trout. PhD. Dissertation, Idaho State
University.
Pough, F. H., E. M. Smith, D. H. Rhodes, and A. Collazo. 1987. The abundance of salamanders
in forest stands with different histories of disturbance. Forest Ecology and Management
20:1-9.
Powers, P. D., K. Bates, T. Burns, B Gowen, and R. Whitney. 1997. Culvert hydraulics related
to upstream juvenile salmon passage. Washington Department of Fish and Wildlife.
Olympia, WA.
30
Pyle, R. E., W. W. Beverage, T. Yoakum, D. P. Amick, W. F. Hatfield, and D. E. McKinney.
1982. Soil survey of Randolph County area, main part, West Virginia. United States
Department of Agriculture, Soil Conservation Service and Forest Service.
Reed, R. A., J. Johnson-Barbard, and W. L. Baker. 1996. Contribution of roads to forest
fragmantation in the Rocky Mountains. Conservation Biology 10:1098-1106.
Reid, L. M. and T. Dunne. 1984. Sediment production from forest road surfaces. Water
Resources Research 20:1753-1761.
Resetarits, Jr., W. J. 1995. Competitive asymmetry and coexistence in size-structured
populations of brook trout and spring salamanders. Oikos 73:188-198.
Resetarits, Jr., W. J. 1991. Ecological interactions among predators in experimental stream
communities. Ecology 72:1782-1793.
Richardson, J.S. 2002. KEYNOTE: Species at risk - amphibians and other life on the edge in
British Columbia in The New Millennium: can threatened / endangered species recover?
Proceedings f the 40th Annual Meeting of the Canadian Society of Environmental
Biologists. Toronto, Ontario, Canada.
Rocco, G. L., R. P. Brooks, and J. T. Hite. 2004. Stream plethodontid assemblage response
(SPAR) index: development, application, and verification in the MAHA final report.
Penn State Cooperative Wetlands Center, Penn State Institutes of the
Environmental/Department of Geography, Pennsylvania State University, University
Park, PA. Report No. 2004-1.
Ross, B., T. Fredericksen, E. Ross, W. Hoffman, M. L. Morrison, J. Beyea, M. B. Lester, B. N.
Johnson, and N. J. Fredericksen. 2000. Relative abundance and species richness of
herpetofauna in forest stands in Pennsylvania. Forest Science 46:139-146.
31
Semlitsch, R. D., D. E. Scott, and J. H. K. Pechmann. 1988. Time and size at metamorphosis
related to adult fitness in Ambystoma talpoideum. Ecology 69:184-192.
Sih, A. and D. E. Wooster. 1994. Prey behavior, prey dispersal, and predator impacts on stream
prey. Ecology 75:1199-1207.
Smith, G. R., J. Rettig, G. G. Mittelbach, J. L. Vallulis, and S. R. Schaack. 1999. The effects of
fish on assemblages of amphibians in ponds: a field experiment. Freshwater Biology
41:829-837.
Spight, T. M. 1967. Population structure and biomass production by a stream salamander. The
American Midland Naturalist 78:437-447.
Spurgeon, A. B. 2002. Comparison of herpetofaunal species composition and response to edge
on the Camp Dawson Collective Training Area, Preston County, West Virginia. M.S.
Thesis, West Virginia University, Morgantown, West Virginia, USA.
Taylor, R. N. and M. Love. 2003. Fish passage evaluation at stream crossings. California
salmonid stream habitat restoration manual.
Thompson, P. D. and F. J. Rahel. 1998. Evaluation of artificial barriers in small Rocky
Mountain streams for preventing the upstream movement of brook trout. North
American Journal of Fisheries Management 18:206-210.
Tyler, T. J., W. J. Liss, R. L. Hoffman, and L. M. Ganio. 1998. Experimental analysis of trout
effects on survival, growth, and habitat use of two species of Ambystomatid salamanders.
Journal of Herpetology 32:345-349.
Warren, Jr., M. L. and M. G. Pardew. 1998. Road crossings as barriers to small-stream fish
movement. Transactions of the American Fisheries Society 127:637-644.
32
West Virginia Department of Environmental Protection. 2003. West Virginia 2002 303 (d)
impaired streams listing—final. Division of Water and Waste Management.
West Virginia Geographic Information System Technical Center. 1999. Digital elevation model
10-meter. West Virginia Geologic and Economic Survey. UTM Zone 17 NAD 27.
West Virginia University Natural Resource Analysis Center, West Virginia Cooperative Fish and
Wildlife Research Unit, West Virginia Division of Natural Resources, and United States
Geologic Survey Biological Resources Division. 2002. West Virginia gap analysis
project. UTM Zone 17 NAD 27.
White, J. A. 2004. Geomorphic analysis of stream crossings in a portion of the upper Cheat
River basin. M.S. Thesis. West Virginia University, Morgantown, West Virginia, USA.
Ziemba, R. E., M. T. Myers, and J. P. Collins. 2000. Foraging under the risk of cannibalism
leads to divergence in body size among tiger salamander larvae. Oecologia 124:225-231.
33
Table 1: Salamander species of interest in the Shavers Fork and Dry Fork watersheds, according
to Green and Pauley (1987).
Family Salamandridae
Red-spotted Newt
Notophthalmus viridescens
Family Plethodontidae
Northern Dusky Salamander
Desmognathus fuscus
Mountain Dusky Salamander
Desmognathus ochrophaeu
Appalachian Seal Salamander
Desmognathus monticola
Northern Spring Salamander
Gyrinophilus porphyriticus
Midland Mud Salamander
Pseudotriton montanus
Northern Red Salamander
Pseudotriton rubber
Northern Two-lined Salamander
Eurycea bislineata
Longtail Salamander
Eurycea longicauda
34
Figure 1. Shavers Fork and Dry Fork watersheds located in Randolph and Tucker County in
eastern West Virginia.
35
Figure 2. The Shavers Fork watershed in Randolph and Tucker County, West Virginia.
36
Figure 3. Land cover type in the Shavers Fork watershed.
37
Figure 4. Culvert sites surveyed in the lower Shavers Fork watershed.
38
Figure 5. Tributary to Gandy Creek at site 420 shows a typical small stream in the Dry Fork
watershed.
39
Figure 6. Gandy Creek and upper Dry Fork watershed located in Randolph County, West
Virginia.
40
Figure 7. Land cover type in Gandy Creek and upper Dry Fork watershed.
41
Figure 8. Culvert sites surveyed in the Gandy Creek and upper Dry Fork watersheds.
42
Figure 9. Salamander sites sampled in the Gandy Creek and upper Dry Fork watershed.
43
Figure 10. Laurel Fork watershed located in Randolph and Tucker County, West Virginia.
44
Figure 11. Land cover type in the Laurel Fork watershed.
45
Figure 12. Glady Fork watershed located in Randolph and Tucker County, West Virginia.
46
Figure 13. Land cover type in the Glady Fork watershed.
47
Figure 14. Salamander sites sampled in the Glady Fork watershed.
48
Figure 15. Red Creek watershed located in Tucker County, West Virginia.
49
Figure 16. Land cover type in the Red Creek watershed.
50
Figure 17. Lower Dry Fork watershed located in Tucker County, West Virginia.
51
Figure 18. Land cover type in lower Dry Fork watershed.
52
Figure 19. Culvert sites surveyed in the lower Dry Fork watershed.
53
Figure 20. Salamander sites sampled in the lower Dry Fork watershed.
54
Figure 21. Upstream of site 420 showing typical shade levels on streams in the study area away
from vegetation breaks caused by the presence of roads.
55
Figure 22. Culverts were often undersized such as this culvert at site 201. Undersized culverts
lead to ponding of water and aggradation at higher flows.
56
Figure 23. Limestone rip rap boulders placed in the stream channel downstream from site 201.
Channelization is common at road crossings.
57
Figure 24. Common types of culverts according to Taylor and Love (2003).
58
Figure 25. Corrugated steel pipe is a common construction material used for culverts.
59
Figure 26. Concrete is a common construction material used for culverts.
60
Figure 27. High flow velocity through culverts commonly prevents the passage of aquatic
organisms.
61
Figure 28. Lack of sufficient water depth commonly prevents the passage of aquatic organisms.
62
Figure 29. Lack of a pool at the culvert outlet commonly prevents the passage of aquatic
organisms.
63
Figure 30. Excessive outlet hang commonly prevents passage of aquatic organisms.
64
Figure 31. The retention of bedload inside culverts creates varied flow velocities and promotes
the passage of aquatic organisms.
65
Figure 32. Brook trout are commonly found in small streams within the study area and are
predators of salamanders.
66
Chapter II:
Culvert Effects on Stream and Stream-side Salamander Habitats in the Dry
Fork and Lower Shavers Fork Watersheds in West Virginia
Ryan L. Ward
James T. Anderson
J. Todd Petty
Division of Forestry
West Virginia University
P.O. Box 6125
Morgantown, WV 26506
J. Steven Kite
Department of Geology and Geography
West Virginia University
P.O. Box 6300
Morgantown, WV 26506
Michael Strager
Natural Resource Analysis Center
West Virginia University
2009 Agricultural Sciences Building
Morgantown, WV 26506
Ronald H. Fortney
Department of Civil and Environmental Engineering
West Virginia University
P.O. Box 6070
Morgantown, WV 26506.
67
Culvert effects on stream and stream-side salamander habitats in the Dry Fork and lower
Shavers Fork Watersheds in West Virginia.
Ryan L. Warda, James T. Andersona, J. Todd Pettya, J. Steven Kiteb, Michael Stragerc, and
Ronald H. Fortneyd
a
Division of Forestry, West Virginia University, P.O. Box 6125, Morgantown, WV 26506
Department of Geology and Geography, West Virginia University, P.O. Box 6300,
Morgantown, WV 26506
c
Natural Resource Analysis Center, West Virginia University, 2009 Agricultural Sciences
Building, Morgantown, WV 26506
d
Department of Civil and Environmental Engineering, West Virginia University, P.O. Box 6070,
Morgantown, WV 26506.
b
Abstract
Road and stream intersections require a crossing that allows safe passage of water and
vehicles. Culverts are normally used when roads cross small streams. Recently, passage of
aquatic organisms through culverts has received increased attention. Geographic information
systems (GIS) analysis was performed in this study to determine the degree of salamander
habitat fragmentation in the lower Shavers Fork and Dry Fork watersheds in Tucker and
Randolph counties in West Virginia. Culverted sites on state roads in the watershed were visited
and salamander barriers were categorized as complete, partial, or nonbarrier, based on outlet
hang and substrate. Complete barriers occurred at 55.0% of the sites visited and partial barriers
at 34.2%. Analyses showed that 20.6% of the total stream length in the Dry Fork watershed and
18.4% in the Shavers Fork watershed were isolated by at least a partial barrier. Outlet hang
height and the presence (or absence) of streambed substrate were the main determinants of
stream salamander passage. Outlet hang was positively correlated with stream gradient and
culvert slope. Culverts containing streambed substrate occurred on lower gradient streams, had
Written in the style of Ecological Engineering.
68
lower culvert slope, and had a greater width compared to culverts lacking substrate. Solutions to
facilitate movement of salamanders and other aquatic organisms are needed to maintain stream
connectivity and provide mitigation opportunities.
Author Keywords: stream salamanders; culverts; habitat fragmentation; roads; streams; passage
Introduction
Roadways are a necessary component of human lives and a prominent feature on the
landscape. The need for roads is not likely to change, and therefore as wildlife managers and
environmental stewards we should strive to minimize their impacts on wildlife and their
ecosystems. Practitioners of stream restoration have begun to develop ways to maintain stable
road crossings and functioning streams (Johnson 2002). However, the field of culvert
installation to maintain geomorphic stability and provide passage of fish and other aquatic
organisms is still in its infancy (Sylte 2002).
Culverts are commonly used for road drainage and stream channels (Adair et al. 2002).
Common types of culverts include circular, open-bottom arch, pipe arch, and box culverts, and
construction materials used include corrugated steel pipe, structural steel pipe, aluminum, plastic,
concrete, and wood (Taylor and Love 2003). In addition to types of crossing structures and
construction materials, consideration of fish and wildlife passage must be given during the
transportation planning process to maintain the ecological integrity of streams (Adair et al.
2002).
The ecological importance of headwater streams has traditionally been underestimated
(Gomi et al. 2002). Headwater streams are important for the breakdown of organic matter,
nutrient transformation, and nutrient retention (Meyer and Wallace 2001). Connectivity of
headwater streams to downstream reaches is important for sediment transportation and
69
recolonization of faunal communities after disturbance (Meyer and Wallace 2001, Gomi et al.
2002).
While single culverts may not significantly affect overall populations of aquatic
organisms, the cumulative effects of culverts in the watershed and across the landscape can be
substantial (Sylte 2002). Gibson et al. (2005) found that 3,000 m2 of benthic stream habitat was
lost due to poor installation of culverts along a 210 km stretch of highway in Canada. In the
United States, of 8.4 million km of river (based on 1:500,000 scale maps) only 2% of the total
length is free flowing for at least 200 km (Meyer and Wallace 2001).
Salamanders have received little consideration for passage through culverts (Sylte 2002).
Most studies focus on the passage of fish, including the development of computer software to
analyze culverts (Furniss et al. 2000, Sylte 2002). Excessive outlet hang of culverts is a common
condition that blocks fish passage (Taylor and Love 2003, Sylte 2002). The culvert outlet
bottom should be below the outlet pool to prevent hanging barriers (Fitch 1995). However, fish
are able to use the outlet pool to swim and attain a sufficient velocity to overcome modest outlet
hang (McClellan 1970, Lauman 1976, Powers 1984, Blevins and Carlson 1988). Amphibians
are weak swimmers compared to fish (Jackson 2003). The reduced swimming ability of
salamanders would most likely inhibit their ability to make use of outlet pools as areas to build
up speed for jumping out of the water although salamanders on land can jump short distances to
avoid predators (Green and Pauley 1987).
D’Aout and Aerts (1997) found that adult axolotls (Ambystoma mexicanum) had lower
swimming ability than most fishes. Lower swimming abilities suggest that high water velocities
through culverts would have a greater impact on salamanders than on fish. During normal runoff
conditions, water velocities in streams are typically 0-0.9 m/sec near the channel bed and stream
70
margins where most aquatic organisms live and travel (Sylte 2002). Differences in water
velocity in culverts compared to natural channels are likely to impede salamander passage
(Jackson 2003). Flow velocities in culverts with gradients as low as 1-2% may exceed 1.2-1.5
m/sec under normal runoff conditions without even constricting the channel width (Sylte 2002)
Velocities in culverts exceed extreme velocities in the natural channel that average 0.9-1.8 m/sec
during bankfull flows (Sylte 2002). Terrestrial capabilities of salamanders could possibly allow
them to navigate through or around culverts, but leaving their preferred habitat and crossing
roads may make them vulnerable to predators and automobile mishaps (Jackson 2003).
Ideally culvert bottoms should mimic natural streambeds (Baker and Votapka 1990). The
physical nature of the bed material in a stream may be the most biologically significant factor
affecting stream functions (Cummins 1974). Salamanders use the channel substrate for refuge
and foraging (Moore et al. 2001). Road networks affect peak flows and may trigger or stop
debris flows that determine the bed material of a stream (Jones et al. 2000, Kish 2004). The
presence of streambed material inside culverts causes variability in velocities that can allow
passage of aquatic organisms (Kahler and Quinn 1998). The lack of bed material in culverts has
been described positively because the culvert is self cleaning and less likely to clog (Johnson
2002). However, this view shows a lack of consideration for the passage of aquatic organisms.
Installation of bottomless or embedded culverts has risen in recent years. Bottomless culverts
simulate natural stream conditions and promote passage of aquatic organisms (Adair et al. 2002).
However problems with bottomless culverts include a lack of hydraulic efficiency, lower
structural integrity, and propensity for the undermining of roadway foundations (McClellan
1970, Lauman 1976). The footings of bottomless culverts are easily undermined on streams with
steep gradients, highly unstable streambeds, or where the stream gradient changes within reach
71
of the culvert (Baker and Votapka 1990). The cost of bottomless culverts is substantially more
than regular culverts and they take longer to install (Baker and Votapka 1990). Embedding a
culvert allows for a stream channel to form inside the culvert, mimicking the rest of the stream,
and allowing a sufficient depth of water to facilitate aquatic organism passage (Taylor and Love
2003). Embedding a culvert does not guarantee the retention of bed material and subsequent
flows after installation may clear all substrate from the culvert. The variation in velocities and
similarity to natural stream channels created by substrate in the culvert is important for
salamander passage (Jackson 2003).
The objectives of our study were to:
1. Determine the extent of habitat fragmentation by roads and culverts in the lower Shavers
Fork watershed and Dry Fork watershed, West Virginia.
2. Determine installation parameters of corrugated steel pipe culverts that will prevent
excessive outlet hang and best promote the passage of stream and stream-side
salamanders.
3. Determine installation parameters of circular and pipe arch culverts constructed of
corrugated steel pipe culverts that will allow for the retention of bedload material and
best promote the passage of stream and stream-side salamanders.
We hypothesized that a significant portion of small streams in both the Shavers Fork
watershed and the Dry Fork watershed are isolated from the mainstem of each river by complete
and partial barrier culverts. We also hypothesized that culverts lacking severe outlet hang will
have a lower culvert slope, be shorter, and occur on streams with low gradients. Finally we
hypothesized that culverts retaining bedload will have a lower culvert slope, greater width, and
occur on streams with low gradients.
72
Study Area
We conducted culvert surveys in the lower Shavers Fork and the Dry Fork watersheds of
the upper Cheat River basin (Figure 1). The study area was located in Randolph and Tucker
counties in eastern West Virginia.
Shavers Fork flowed to the town of Parsons, West Virginia, where it combined with the
Black Fork to form the Cheat River. Shavers Fork was a cold and cool water fishery with most
tributaries consisting of high gradient mountain streams. Surveys were conducted in the lower
portion of the watershed, which contained an extensive state road network. The U.S. Forest
Service or private landowners owned most roads in the upper portion of the watershed, and this
area was excluded for this reason.
The Dry Fork flowed near Parsons where it combined with the Blackwater River to form
the Black Fork. Dry Fork maintained a cool-water fishery and a cold-water fishery in some
sections. Major tributaries to Dry Fork included Gandy Creek, Laurel Fork, Glady Fork, and
Red Creek. Laurel Fork and Glady Fork were excluded from the study due to a lack of state
owned roads within their sub-watersheds. Red Creek was excluded from the study due to poor
water quality. High gradient mountain stream characterized minor tributaries to Dry Fork and
Gandy Creek. Otter Creek empties into lower Dry Fork from the Otter Creek Wilderness area.
The average winter temperature in the study area was –0.5 ûC and the average summer
temperature was 20.1 ûC (Losche and Beverage 1967, Pyle et al. 1982). Average annual rainfall
in the study area was 116 cm (Losche and Beverage 1967, Pyle et al. 1982). Prevailing winds in
the study area occur from the northwest and west (Losche and Beverage 1967, Pyle et al. 1982).
Elevations ranged from 518 m to 1,472 m (West Virginia Geographic Information
System Technical Center 1999). The most surface abundant geologic map units were the
73
Pottsville group, Mauch Chunk group, Hampshire formation, and Chemung group (Cardwell et
al 1968). Major soil associations were the Dekalb-Buchanan association, Calvin-high base
substratum-Belmont-Meckesville association, Dekalb-Calvin-Belmont association, Gilpin
association, Barbour-Pope-Sequatchie association, Calvin association, Dekalb-Berks-Calvin
association, Dekalb-Gilpin association, very stony land-Ernest-Brinkerton-Leetonia association,
and the very stony land-Dekalb association (Losche and Beverage 1967, Pyle et al. 1982).
Methods
Our study was conducted in three phases. First, we used geographic information system
(GIS) analysis to create working maps for field crews. Next, we visited each culvert site and
performed field surveys. Last we conducted data analyses to interpret collected data.
GIS
We used ArcMap GIS ver. 8.2 software from Environmental Systems Research Institute
(2002) to determine the location of streams that drain at least 40.5 ha (100 acres). Then we
added a layer of roads to find stream and road intersections. We used these layers to construct
maps to assist field crews in finding stream crossings. Streams listed as impaired by the West
Virginia Department of Environmental Protection (2003) according to the Clean Water Act
Section 303d were excluded. Streams with drainage areas less 40.5 ha, although ecologically
important, were excluded because their small size offers a lack of any significant mitigation
opportunities.
Culvert Surveys
We conducted culvert surveys in June-October of 2003. We visited each site to
determine the crossing type (culvert, bridge, or ford). Only sites with culverts were surveyed.
We recorded the type of each culvert, construction materials, length, and diameter or height and
74
width (Love 2000). The active channel of the stream was defined as the portion of the channel
that was lacking vegetation due to frequent water flows (Taylor and Love 2003). We took 4
measurements of the active channel width upstream of the culvert, and we took 4 measurements
of bankfull width and bankfull depth at a typical riffle. We determined the length of stream
reaches to be surveyed by multiplying the mean active channel width by 30. We used a
minimum length of 30 m and a maximum length of 100 m to limit reach lengths. The culvert
generally was located in the center of each stream study reach. Where a culvert occurred near a
stream mouth, we surveyed the entire downstream reach if it was not long enough to meet the
total desired reach length.
We surveyed the longitudinal profile of the stream in the study reach from the head of the
first riffle to the head of the last riffle. Following the protocols of Taylor and Love (2003), we
recorded elevations at the following points: head of upstream riffle, inlet, outlet, deepest part
within 2 m of outlet, deepest part of outlet pool, tailwater control, active channel margin at the
tailwater control, head of downstream riffle, and additional slope breaks. We collected
additional habitat measurements, including estimation of the percent canopy of the overstory,
shrub, and herbaceous layers for each bank both upstream and downstream, presence or absence
of continuous suitable salamander (>40 mm) substrate above, below, and inside culverts, and
completion of Environmental Protection Agency (EPA) Habitat Assessment forms (Barbour et
al. 1999) for up and downstream.
Data Analysis
We conducted surveys on all culverts on state roads and the full data set was used to
analyze the barrier effects as a whole on the landscape. We considered any culvert with an outlet
hang over 0.10 m a complete barrier for salamanders. Outlet hangs of 0.05-0.10 m were
75
considered partial barriers. Culverts with outlet hangs under 0.05 m but lacking continuous
substrate were considered partial barriers. Only if a culvert had an outlet hang under 0.05 m and
continuous substrate, did we consider it passable (Figure 2). We used ArcMap GIS ver. 9.0
software from Environmental Systems Research Institute (2004) to determine the amount of
stream reaches fragmented by complete and partial barrier culverts. We excluded Otter Creek
from analysis of fragmentation because it is wilderness area and lacks and active road system.
We used a G-test of association to test if the barrier types observed differed from expected values
and for comparison of barrier types between watersheds (Sokal and Rohlf 1994).
Corrugated steel pipe was the most common construction material used for culverts in the
study area. Because of their numbers, analyses of culvert parameters were restricted to culverts
constructed of corrugated steel pipe. Some stream crossings contained multiple culvert barrels.
Because of the low number of sites with multiple barrels, they were removed from analyses of
culvert parameters. We used linear regression to examine relationships between hang height and
stream gradient, culvert length, and culvert slope. We conducted analyses of bedload retention
only on sites with continuous substrate upstream of the culvert. This restriction was intended to
rule out bedrock streams that were bedload limited. We used a G-test to compare circular and
pipe arch culverts, and their ability to retain stream bedload (Sokal and Rohlf 1994). We used ttests assuming equal variances to compare culvert variables for culverts with and without
continuous substrate. We analyzed the variables of circular culverts separately from pipe arch
culverts because different shapes likely influenced hydraulic forces. Comparisons we conducted
included culvert variables (slope and diameter/width), stream variables (gradient), and ratios of
culvert variables versus stream variables (culvert diameter/width versus active channel width,
76
culvert slope versus stream gradient, and culvert cross-sectional area versus bankfull crosssectional area).
Results
Culvert Surveys
A total of 120 culvert sites was surveyed in the Dry Fork (n = 68) and Shavers Fork (n =
52) watersheds. Single culverts occurred at 116 sites, while the remaining 4 sites had double
culverts. Circular culverts were the most common type, occurring at 66 sites. Pipe arch culverts
were the second most abundant type, occurring at 36 sites. Box culverts occurred at 13 sites.
Combinations of box and circular culverts occurred at 5 sites. These combined culverts
consisted of old stone box culverts that were lengthened with corrugated steel or concrete pipe
culverts when the roadway was widened. Corrugated steel pipe was the most common
construction material occurring at 94 sites, and concrete was used at 20 sites. The remaining 6
sites were made of stone or stone and corrugated steel pipe combintations.
Habitat Fragmentation
Culverts were likely to create barriers to salamanders as salamander movement
conditions were classified as complete barriers at 55.0%, partial barriers at 34.2%, and
unrestricted passage at 10.8% of culverts surveyed (Figure 3; Appendix 1) (n = 120, G2 = 38.90,
P < 0.001). Culverts in the Shavers Fork watershed were more likely to be complete barriers
than in the Dry Fork watershed (Figure 3) (G2 = 14.32, P < 0.001). Barriers isolated 20.6% of
the total lengths of stream draining >40.5 ha in the Dry Fork watershed and 17.4% in the Shavers
Fork watershed (Table 1; Appendix 2).
77
Outlet Hang
Hang height was correlated with stream gradient for corrugated steel pipe culverts (n =
90, R2 = 0.185, P < 0.001) (Figure 4). A correlation also existed between hang height and culvert
slope (R2 = 0.096, P = 0.002) (Figure 5). A weak correlation was found between culvert length
and hang height (R2 = 0.056, P = 0.02) (Figure 6).
Continuous Substrate
Of the 120 culverts surveyed, upstream reaches had continuous substrate in 87.3% of
surveyed sites, and downstream reaches had continuous substrate in 85.6% of the sites. Only
17.9% of the sites had continuous substrate throughout the entire culvert length (Figure 7).
Culverts tended to create breaks in the stream channel material (n = 120, G1 =61.49, P < 0.001).
We performed analyses on a total of 53 circular culverts (9 with continuous substrate)
and 29 pipe arch culverts (5 with continuous substrate). No difference was found between the
proportion of circular culverts with substrate (17.0%) and the proportion of pipe arches with
substrate (17.2%) (n = 82, G1 = 0.31, P = 0.54).
Culvert slopes were lower for culverts with continuous substrate compared to those
lacking substrate (Table 2). Culvert diameters were greater for culverts with continuous
substrate compared to those without continuous substrate (Table 2). Culverts with continuous
substrate occurred on streams with significantly less gradient than culverts without continuous
substrate (Table 2). There was no difference for stream gradient of pipe arches, but pipe arches
with continuous substrate were wider than culverts lacking continuous substrate (Table 3).
Discussion
The majority of culverts surveyed were complete or partial barriers to stream salamanders
(89%). These sites most likely prevented the movement of salamanders at all or most flow
78
conditions. Complete barriers occurred at a greater frequency than expected from chance alone.
A higher percentage of culverts in the lower Shavers Fork watershed were complete barriers
compared to the Dry Fork watershed. One possible explanation for this result is the number of
high gradient streams in the lower Shavers Fork watershed because of steeper topography. The
Dry Fork watershed had more low gradient streams, especially in the Gandy Creek
subwatershed. Extra consideration must be given in placing culverts on high gradient streams
(>8%). Upstream movement of salamanders and connectivity of salamander habitats are
important for the persistence of populations over time (Lowe and Bolger 2002, Lowe 2003).
Culverts isolated headwater streams from downstream areas. Headwater streams are the
most important habitat for stream salamanders and have the highest densities of salamanders
(Resetarits 1995, Ohio EPA 2002). In the Dry Fork watershed, 20.6% of the total stream length
was isolated from colonization sources. In the Shavers Fork watershed, 18.4% of the total
stream length was isolated. Isolated streams consisted of low order headwater streams separated
from the mainstem of each river. Salamanders move into these areas to exploit resources
available in less hydraulically stable streams lacking fish populations (Jackson 2003).
This study was only conducted on state owned roads. The study area also has a plethora
of Forest Service and private roads. The culverts on these roads greatly increase the amount of
habitat fragmentation occurring along streams in the two watersheds. Streams with drainage
areas under 40.5 ha were not measured in the study. These smaller streams would contribute
greatly to salamander habitats, and add to the total amount of isolated habitat.
Outlet Hang
Corrugated steel pipe was the most common construction material used by the West
Virginia Division of Highways. Single culverts made of corrugated steel pipe occurred at 82%
79
of the sites visited. Corrugated steel pipe was used extensively, probably due to its relative low
cost compared to other materials and ease of installation relative to concrete and stonework
(Baker and Votapka 1990).
Higher gradient streams were more likely to have a culvert with a perched outlet. High
gradient streams also require culverts to be installed at a steeper gradient. Stream gradient better
explained the total hang height when compared to culvert slope, but culvert lengths explained
little about hang height. Most culverts were approximately the same length for each road
depending on the number of vehicle lanes. Culvert lengths were influenced by standard stock
lengths and did not vary as much as hang heights.
While large drops and small waterfalls are common in high gradient streams, the outlet
hang from a culvert creates a substantial barrier because the road and associated fill prevent the
formation of side channels and prevent organisms from using the stream banks to overcome a
small part of a stream where hydraulic forces exceed their swimming abilities. Salamanders can
normally walk around a small waterfall, but to overcome the drop from a culvert barrel they must
climb the roadfill, cross the road surface (usually at least 8 m), and descend back down to the
stream channel. If salamanders are unable to enter the culvert barrel, the road is most likely a
barrier to most salamander passage.
Culverts with hang heights of ≥0.10 m were considered complete barriers to stream
salamanders. We hypothesized that small body sizes and weak swimming abilities probably
prevent stream salamanders from overcoming large outlet hangs. Fish can overcome larger hang
height because of their ability to use the outlet pool to gain speed for jumping (McClellan 1970,
Lauman 1976, Powers 1984, Blevins and Carlson 1988). Culverts with hang heights of 0.050.10 m were considered partial barriers. These hang heights were only estimates we made for
80
the passage of adult stream salamanders. We feel confident that stream salamanders cannot
overcome hang heights ≥0.10 m, but lower heights possibly might be overcome if the right
conditions exsited. Passage for larval salamanders would likely still be prevented at any outlet
hang height. Older life stages of salamanders are more likely to undergo upstream movements
(Bruce 1986). However, many salamander species have larval stages longer than a year (Green
and Pauley 1987), and salamanders often undergo upstream movement to find suitable wintering
locations (Ashton 1975, Ashton and Ashton 1978). Hang heights under 0.05 m were considered
passable even though larval stream salamanders would not be likely to overcome any hang at the
outlet. We hypothesized that adult salamanders could enter a culvert barrel with hang heights in
this range.
Culverts that prevent up and downstream movements of salamanders affect the structure
of populations and the ability of individuals to locate wintering sites (Ashton 1975, Ashton and
Ashton 1978, Bruce 1986). Also, some salamanders use streambeds as movement corridors
(Gibbs 1998a). Workers should try to minimize outlet hang in crossing structures. This
minimization may require the building of a bridge on a high gradient stream and avoiding the use
of a culvert altogether.
Continuous Substrate
The presence of continuous streambed substrate throughout the culvert was required to
categorize a culvert as a nonbarrier. Streambed material creates variations in the flow velocity
that allow salamanders to move upstream (Jackson 2003). To pass a culvert the salamander had
to be able to enter the culvert barrel and then negotiate the entire culvert length. Retention of
streambed material allows salamanders to overcome high velocities found inside culverts.
81
Culverts in the study area are serving poorly to retain bed material and prevent breaks in
the substrate. Only 17.9 % of the sites visited contained continuous substrate through the culvert
while most upstream reaches (87.3 %) and downstream reaches (85.6 %) had continuous
substrate. These results show a failure of culverts to mimic natural channels and show a need for
better culverts and installation methods. If dimensions and placement of corrugated steel pipe
culverts can be done properly to prevent breaks in the stream substrate, they may provide a low
cost alternative to other options such as bridges and concrete culverts that provide adequate
substrate.
Culvert slope, overall stream gradient, and diameter were significant variables that
affected the ability of circular culverts to retain bed material and prevent breaks in the channel
substrate. Similar patterns existed for pipe arch culverts, although a lack of sample size and high
variance prevented obtaining statistically significant results. Width did emerge as a significant
variable for pipe arches with continuous substrate.
Culvert slope affects flow velocity. Circular culverts with substrate had a lower mean
slope than culverts lacking substrate. Culverts with lower slopes likely had lower water velocity
that would help with the retention of stream bedload. Sites with continuous substrate had a
lower mean stream gradient compared to sites without continuous substrate. In low gradient
streams, water velocities in culverts are probably more similar to velocities in the natural
channel. High gradient streams require steeper culverts. In these streams a standard corrugated
steel pipe culvert may not suffice and additional modifications may be required to lower flow
velocity and allow deposition. High gradient streams often form a series of step pools and a
crossing structure that mimicked this step pool configuration might better promote the passage of
aquatic organisms.
82
Circular culverts with continuous substrate had larger diameters than culverts lacking
continuous substrate. A similar trend was observed in the width of pipe arches. A larger
diameter or width prevents pooling at the inlet of a culvert and subsequent deposition of bed
material before entering the culvert (Sylte 2002). White (2004) found that 91% of aggraded
reaches at culverts within the study area were at least partially caused by low conveyance. Small
diameters or widths also constrict the flow of streams, which can cause increased water velocity
(Sylte 2002). Wider culverts better simulate natural stream conditions (Sylte 2002). Larger
culverts also prevent the failure of fill dirt used in the roadbed that can be a source of
sedimentation in streams (Saltzman and Koski 1971). Wide culverts tend to be found on larger
streams, and these large streams tend to be lower in gradient and able to successfully retain
substrate. More attention should be given to smaller streams because of their overall importance
and the historical lack of consideration given to them.
No difference was found in the proportions of circular versus pipe arch culverts that had
continuous substrate. Pipe arch culverts are often used to limit the amount of roadfill needed, but
still provide enough cross-sectional area for the passage of stream flows (Baker and Votapka
1990). Due to the increased width at the bottom of pipe arch culverts, a larger sample size of
pipe arches may show a higher proportion with continuous substrate. However, the slope and
low roughness factor of both circular and pipe arch culverts are the most important determinants
of water velocity (Lauman 1976, Blevins and Carlson 1988, Fitch 1995). The effects of slope
and roughness on water velocity may affect the retention of bedload more than culvert width.
Data analyses on pipe arches often failed to reach a significant alpha level of 0.05. All
pipe arches were sampled within the study area, so culverts in additional watersheds need to be
surveyed to increase the sample size. An increased sample size would likely result in more
83
statistical difference. Streams are highly variable which makes each culvert site different. Many
factors at each site could influence the ability of the culvert to retain continuous substrate. Large
woody debris caught inside a culvert will slow water velocities and trap bedload in the culvert.
Debris jams can compromise the hydraulic integrity of a culvert and are not a desirable means to
facilitate bedload retention.
Studies have shown that culverts can be modified to benefit aquatic organisms (Blevins
and Carlson 1988). Velocities have been slowed with the addition of corrugations and baffles
inside culverts, which increase roughness (Taylor and Love 2003). Baffles can improve fish
passage and show good durability (McClellan 1970, Blevins and Carlson 1988). Problems with
baffles include high cost, difficulty in fabrication, sedimentation, debris jams, icing, and
increased turbulence through the culvert (Blevins and Carlson 1988, Baker and Votapka 1990,
Fitch 1995). The increased turbulence created by baffles may be negative for salamanders.
However, if baffles trap some sediment this might provide a suitable surface for salamanders to
use during passage. Weirs can be used to adjust stream gradients at the inlet and outlets of
culverts, compensating for large drops and hydraulic forces (Lauman 1976, Taylor and Love
2003). If water is pooled into the culvert outlet this would help salamanders enter culverts.
Conclusions and Management Implications
Economical solutions are needed for highway construction, but not at the expense of
biological communities. Isolated populations are vulnerable to inbreeding depression and
demographic and environmental stochasticity (Mills and Smouse 1994). Connectivity of
populations is important for genetic interchange, movement in response to environmental
changes, and recolonization of locally extirpated populations (Beier and Loe 1992). Fragmented
populations of salamanders have different levels of genetic diversity compared to continuous
84
populations (Gibbs 1998b). While single culverts may not affect overall populations, the
cumulative effect of the entire road network can be substantial. In the study area 55.0% of
culverts were complete barriers and 34.2% were partial barriers. The stream fragmentation
effects in the Dry Fork and lower Shavers Fork watersheds caused by state road culverts were
20.6% and 18.4% respectively of the total stream length. This estimate of fragmentation is
conservative because Forest Service and private roads were not included in the study.
Consideration should be given at each stream crossing for biological communities and the
ecological integrity of the stream. If passage for both strong and weak swimming organisms is
considered, a wide range of organisms will benefit.
Excessive outlet hang was a problem at most culverts. When crossing high gradient
streams requiring steeply sloping culverts, other structures should be considered such as bridges
or fords. Weirs can be constructed to raise the elevation of the outlet pool surface into the
culvert barrel. Preventing outlet hang should be a priority on all culverts installed.
Proper dimensions and placement of culverts can reduce the tendency to create breaks in
the channel substrate. Appropriate substrate in culverts is suggested for use by other
herpetofauna as well (Aresco 2005). Standard corrugated steel pipe culverts can be used in low
gradient streams and not create a break in channel substrate. Corrugated steel pipe culverts are
the most commonly used because of relative low cost and ease of installation.
Concrete is not
needed with corrugated steel pipe as opposed to bottomless arches and bridges. Wide culverts
allow the water passing through to better simulate conditions in the natural channel and allow for
the retention of bed material that benefits salamanders and other aquatic organisms. Culverts
that are wider than the stream channel allow for dry areas that would facilitate the passage of
other riparian and upland wildlife (Forman et al. 2003).
85
Low roughness inside culverts allows 2-4 times the flow to pass as an equal section of
natural channel with equal slope (Bell 1973). Increasing the roughness of culverts slows water
velocities and would promote the passage of aquatic organisms. The insides of corrugated steel
pipe culverts are sometimes coated with asphalt coverings to prolong the life of the culvert
(Baker and Votapka 1990). These coatings smooth out corrugations, decreasing roughness, and
affecting small fish that depend on the corrugations as resting bays during passage (Baker and
Votapka 1990). This practice was observed in some culverts in our study area, and likely
reduces the ability of salamanders to pass through culverts.
Our study shows that many mitigation opportunities exist in the present road network.
Regulatory agencies should allow mitigation credit for the replacement of culverts that fail to
allow salamander passage and negatively affect salamander habitat. Installing a culvert where
hydraulic conditions inside are equal to conditions in the natural stream channel should be the
goal of culvert designers. With consideration and planning, humans can lessen the impact that
roads have on wildlife and the environment. The prevention of passage barriers to stream
salamanders will help prevent the isolation of populations and help maintain overall biodiversity
in stream systems. Mitigation credits can be based not only on effects to salamanders but also
fish and other stream fauna.
Most research on culverts and barriers has concentrated on fish. Swimming abilities of
salamanders is relatively unknown compared to most fish species. More data on the abilities of
salamanders would better allow for the analysis of barriers. Future research needed includes the
effects of fragmentation on stream salamander populations and population genetics. Also more
studies are needed on the movement of stream salamanders. Detailed studies are needed on
distances moved, reasons for movement, and timing of movement for different life stages of
86
salamanders. Research on culvert designs should include ways to increase roughness and
preventing increased water velocities. New types of culvert are needed to accommodate passage
of a wider range of aquatic organisms. When designing new culverts, consideration should be
given to cost and ease of installation. Development of methods to retrofit existing culverts to
eliminate passage problems would provide a more cost effective alternative to replacement of
culverts that are still structurally sound. Further research on the ecology of streams is important
to understand the importance that each species plays in the ecosystem function, and how to best
maintain a diversity of species and functions in a world where human-wildlife interactions are
ever increasing.
Acknowledgements
Financial support for the study was provided by the West Virginia Division of Highways
and West Virginia University Division of Forestry. Field assistance was provided by Ira PoplarJeffers, Pat Kish, and Josh White. This is scientific article No. XXXX of the West Virginia
University Agricultural and Forestry Experiment Station.
Literature Cited
Adair, S., M. L. Dereske, J. Doyle, A. Edwards, S. Jacobson, R. Jemison, L. Lewis, W. Melgin.
C. Napper, T. Ratcliff, and T. Warhol. 2002. Management and techniques for riparian
restorations. United States Department of Agriculture, Forest Service, Rocky Mountain
Research Center. Research Note No RMRS-CTR-102.
Aresco, M. J. 2005. Mitigation measures to reduce highway mortality of turtles and other
herpetofauna at a north Florida lake. Journal of Wildlife Management 69:549-560.
Ashton, R. E. 1975. A study of movement, home range, and winter behavior of Desmognathus
fuscus (Rafinesque). Journal of Herpetology 9:85-91.
87
Ashton, R. E. and P. S. Ashton. 1978. Movements and winter behavior of Eurycea bislineata
(Amphibia, Urodela, Plethodontidae). Journal of Herpetology 12:295-298.
Baker, C. O. and F. E. Votapka. 1990. Fish passage through culverts. Report No. FHWA-FL90-006. United States Department of Agriculture, Forest Service Technology and
Development Center, San Dimas, CA.
Barbour, M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid bioassessment
protocols for use in streams and wadeable rivers: periphyton, benthic macroinvertebrates
and fish, second edition. EPA 841-B-99-002. U. S. Environmental Protection Agency;
Office of Water; Washington D. C.
Beier, P. and S. Loe. 1992. A checklist for evaluating impacts to wildlife movement corridors.
Wildlife Society Bulletin 20:434-440.
Bell, M. C. 1973. Fisheries handbook of engineering requirements and biological criteria.
Fisheries-Engineering Research Program, Corps of Engineers, North Pacific Division,
Portland, OR.
Blevins, V. and R. F. Carlson. 1988. Retrofit design of drainage structures for improved fish
passage: literature review. Report No. AK-RD-89-02. Water Research Center, Institute
of Northern Engineering, University of Alaska Fairbanks, Fairbanks, AK.
Bruce, R. C. 1986. Upstream and downstream movements of Eurycea bislineata and other
salamanders in a southern Appalachian steam. Herpetologica 42:149-155.
Cardwell, D. H., R. B. Erwin, and H. P. Woodward. 1968. Geologic map of West Virginia.
West Virginia Geological and Economic Survey. 1:250,000 scale. West Virginia
Department of Environmental Protection and United States Geologic Survey Water
Resources Division. 1998 (digitized). UTM Zone 17 NAD 27.
88
Cummins, K. W. 1974. Structure and function of stream ecosystems. BioScience 24:631-641.
D’Aout, K. and P. Aerts. 1997. Kinematics and efficiency of steady swimming in adult axolotls
(Ambystoma mexicanum). Journal of Experimental Biology 200:1863-1871.
Environmental Systems Research Institute (ESRI). 2002. ArcGIS, Arview license, ver. 8.2
Redlands, California, USA.
Environmental Systems Research Institute (ESRI). 2004. ArcGIS, Arview license, ver. 9.0
Redlands, California, USA.
Fitch, G. M. 1995. Nonanadromous fish passage in highway culverts. Virginia Transportation
Research Council Report VTRC 96-R6.
Forman, R. T., D. Sperling, J. A. Bissonette, A. P. Clevenger, C. D. Cutshall, V. H. Dale, L.
Fahrig, R. France, C. R. Goldman, K. Heanue, J. A. Jones, F. J. Swanson, T. Turrentine,
and T. C. Winter. 2003. Road Ecology Science and Solutions. Island Press, Washington
D.C., U.S.A.
Furniss, M. J., S. Firor, and M. Love. 2000. FishXing 2.0 software and learning system for the
analysis of fish migration through culverts. Stream Notes July 2000:1-4.
Gibbs, J. P. 1998a. Amphibian movements in response to forest edges, roads, and streambeds in
southern New England. Journal of Wildlife Management 62:584-589.
Gibbs, J. P. 1998b. Genetic structure of redback salamander Plethodon cinereus populations in
continuous and fragmented forests. Biological Conservation 86:77-81.
Gibson, R. J., R. L. Haedrich, and C. M. Wernerheim. 2005. Loss of fish habitat as a
consequence of inappropriately constructed stream crossings. Fisheries 30:10-17.
Gomi, T., R. C. Sidle, and J. S. Richardson. 2002. Understanding processes and downstream
linkages of headwater systems. BioScience 52:905-916.
89
Green N. B. and T. K. Pauley. 1987. Amphibians and reptiles in West Virginia. University of
Pittsburgh Press, Pittsburgh, PA, USA.
Jackson, S. D. 2003. Ecological considerations in the design of river and stream crossings in C.
L. Irwin, P. Garrett, and K. P. McDermott, editors, 2003 Proceedings of the International
Conference on Ecology and Transportation. Center for Transportation and the
Environment, North Carolina State University, Raleigh, NC.
Johnson, P. A. 2002. Incorporating road crossings into stream and river restoration projects.
Ecological Restoration 20:270-277.
Jones, J. A., F. J. Swanson, B. C. Wemple, and K. U. Snyder. 2000. Effects of roads on
hydrology, geomorphology, and disturbance patches in stream networks. Conservation
Biology 14:76-85.
Kahler, T. H. and T. P. Quinn. 1998. Juvenile and resident salmonid movement and passage
through culverts. Final Research Report, Research project T9903, Task 96. Washington
State Transportation Center.
Kish, M. P. 2004. Factors controlling landslide initiation as a result of July 2001 high
precipitation events in a section of the Lower New River Gorge, West Virginia. M.S.
Thesis, West Virginia University, Morgantown, WV.
Lauman, J. E. 1976. Salmonid passage at stream-road crossings. A report with department
standards for passage of salmonids. Environmental Management Section, Department of
Fish and Wildlife, Portland, OR.
Losche, C. K., and W. W. Beverage. 1967. Soil survey of Tucker County and part of northern
Randolph County, West Virginia. United States Department of Agriculture, Soil
Conservation Service and Forest Service.
90
Love, M. 2000. Field guide: explanations and instructions for inventory and assessment of
culverts on fish bearing streams. Six Rivers National Forest Watershed Interactions
Team. FishXing Project.
Lowe, W. H. 2003. Linking dispersal to local population dynamics: a case study using a
headwater salamander system. Ecology 84:2145-2154.
Lowe, W. H. and D. T. Bolger. 2002. Local and landscape predictors of salamander abundance
in New Hampshire headwater streams. Conservation Biology 16:183-193.
McClellan, T. J. 1970. Fish passage through highway culverts. United States Department of
Transportation, Federal Highway Administration and Oregon State Game Commission.
Portland, OR.
Meyer, J. L. and J. B. Wallace. 2001. Lost linkages and lotic ecology: rediscovering small
streams in M.C Press, N.J. Huntly, and S. Levin, editors, Ecology: Achievement and
Challenge. The 41st Symposium of the British Ecological Society jointly sponsored by
the Ecological Society of America. Orlando, Florida, USA.
Mills, L. S. and P. E. Smouse. 1994. Demographic consequences of inbreeding in remnant
populations. The American Naturalist 144:412-431.
Moore, A. L., C. E. Williams, T. H. Martin, and W. J. Moriarity. 2001. Influence of season,
geomorphic surface, and cover item on capture, size, and weight, of Desmognathus
ochrophaeus and Plethodon cinereus in Allegheny Plateau riparian forests. The
American Midland Naturalist 145:39-45.
Ohio EPA. 2002. Technical report: Ohio’s primary headwater streams—fish and amphibian
assemblages. Ohio EPA, Division of Surface Water, Columbus, Ohio, USA.
91
Powers, P. D. 1984. Analysis of barriers to upstream fish migration. M.S. Thesis, Washington
State University.
Pyle, R. E., W. W. Beverage, T. Yoakum, D. P. Amick, W. F. Hatfield, and D. E. McKinney.
1982. Soil survey of Randolph County area, main part, West Virginia. United States
Department of Agriculture, Soil Conservation Service and Forest Service.
Resetarits, Jr., W. J. 1995. Competitive asymmetry and coexistence in size-structured
populations of brook trout and spring salamanders. Oikos 73:188-198.
Saltzman, W. and R. O. Koski. 1971. Fish passage through culverts. Oregon State Game
Commission Special Report. Portland, OR.
Sokal, R. R. and F. J. Rohlf. 1994. Biometry: the principles and practices of statistics in
biological research, 3rd edition. Freeman New York.
Sylte, T. L. 2002. Providing for stream function and aquatic organism passage: an
interdisciplinary design. Stream Notes January 2002:1-4.
Taylor, R. N. and M. Love. 2003. Part IX: fish passage evaluation at stream crossings.
California salmonid stream habitat restoration manual.
West Virginia Department of Environmental Protection (WVDEP). 2003. West Virginia 2002
303 (d) impaired streams listing-final. Division of Water and Waste Management.
West Virginia Geographic Information System Technical Center. 1999. Digital elevation model
10-meter. West Virginia Geologic and Economic Survey. UTM Zone 17 NAD 27.
White, J. A. 2004. Geomorphic analysis of stream crossings in a portion of the upper Cheat
River basin. M.S. Thesis, West Virginia University, Morgantown, WV.
92
Table 1
Lengths of stream affected by barrier culverts located on state roads in the Dry Fork and Shavers
Fork watersheds, West Virginia, 2003.
Isolated Stream Segments (km)
Dry Fork
Lower Shavers
Fork
a
Total Stream
Length (km)
Complete
Barriers
Partial
Barriers
Total
341.5a
42.7 (12.5 %)
27.6 (8.1 %)
70.3 (20.6 %)
276
31.9 (11.6 %)
18.7 (6.8 %)
50.6 (17.4 %)
Otter Creek and its tributaries were excluded due to lack of a road system.
93
Table 2
Results of analyses on retention of streambed substrate performed on 53 single, circular culverts
constructed of corrugated steel pipe in the Dry Fork and Shavers Fork watersheds, West
Virginia, 2003.
No Substrate
Variable
Substrate
Mean
SE
Mean
SE
t-test
P-value
Culvert Diameter vs. Active Channel
Width
0.60
0.05
0.66
0.07
-0.61
0.2714
Culvert Slope vs. Stream Gradient
0.69
0.06
0.51
0.09
1.43
0.0797
Bankfull X-sect. vs Culvert X-sect. Area
1.12
0.17
1.20
0.51
-1.12
0.1341
Culvert Slope
6.81
0.58
3.01
0.81
2.84
0.0033
Stream Gradienta
10.86
0.91
6.56
1.46
2.03
0.0240
Culvert Diametera
1.15
0.06
1.87
0.30
-3.86
0.0001
a
a
significant alpha level = 0.05.
94
Table 3
Results of analyses on retention of streambed substrate performed on 29 single, pipe arch
culverts constructed of corrugated steel pipe in the Dry Fork and Shavers Fork watersheds, West
Virginia, 2003.
No
Substrate
Substrate
Mean
SE
Mean
SE
t-test
Pvalue
Culvert Width vs. Active Channel Width
0.93
0.06
1.07
0.13
-0.86
0.1982
Culvert Slope vs. Stream Gradient
0.74
0.08
0.47
0.13
1.37
0.0906
Bankfull X-sect. vs Culvert X-sect. Area
1.70
0.23
1.99
0.29
-0.57
0.2872
Culvert Slope
4.54
0.50
3.18
1.21
1.11
0.1383
Stream Gradient
7.21
0.83
5.79
1.12
0.75
0.2296
Culvert Widtha
1.88
0.13
2.64
0.34
-2.39
0.0120
Variable
a
significant alpha level = 0.05.
95
Figure 1. Map of study area in the lower Shavers Fork and Dry Fork watersheds, West Virginia,
2003. Circles indicate state culverts where surveys were conducted.
96
Culvert (outlet hang height, presence or
absence of streambed substrate)
NO
Outlet hang greater
than 0.10 m
Complete
Barrier
Outlet hang greater
than 0.05 m
YES
Partial
Barrier
YES
NO
NO
Partial
Barrier
Continuous
streambed substrate
through culvert
YES
Nonbarrier
Culvert
Figure 2. Decision tree used to determine barrier status of culverts for stream salamanders in the
lower Shavers Fork and Dry Fork watersheds. Culverts were classified as complete barriers (n =
66), partial barriers (n = 41), and nonbarriers (n = 13).
97
Frequency of Barrier Type
70
Nonbarrier
Partial Barrier
Complete Barrier
60
50
40
30
20
10
0
Shavers Fork
Dry Fork
Total
Figure 3. Frequency of barrier categories for salamanders in the lower Shavers Fork (n = 52) and
Dry Fork (n = 68) watersheds, West Virginia, 2003.
Outlet Hang Height (m)
98
2.00
1.50
1.00
R2=0.185
0.50
0.00
0.00
10.00
20.00
30.00
40.00
Stream Gradient (%)
Figure 4. Linear regression showing the relationship between outlet hang height and stream
gradient for 116 single barrel culverts in the Dry Fork and Shavers Fork watersheds, West
Virginia, 2003.
Outlet Hang Height (m)
99
2.00
1.80
1.60
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
0.00
R2=0.096
5.00
10.00
15.00
20.00
Culvert Slope (% )
Figure 5. Linear regression showing the relationship between outlet hang height and culvert
slope for 116 single barrel culverts in the Dry Fork and Shavers Fork watersheds, West Virginia,
2003.
Outlet Hang Height (m)
100
2.00
1.50
1.00
R2=0.056
0.50
0.00
4.00
9.00
14.00
19.00
24.00
Culvert Length (m)
Figure 6. Linear regression showing the relationship between outlet hang height and culvert
length for 116 single barrel culverts in the Dry Fork and Shavers Fork watersheds, West
Virginia, 2003.
101
100
90
Frequency
80
70
60
Substrate Present
50
40
Substrate Absent
30
20
10
0
Upstream From
Culvert
Downstream From
Culvert
Inside Culvert
Figure 7. Graph showing frequency distribution of areas with continuous substrate at culvert
sites for 120 culverts in the Dry Fork and Shavers Fork watersheds, West Virginia, 2003.
102
Chapter III:
Effects of Road Crossings on Stream and Stream-side Salamander Diversity,
Richness, and Abundance
Ryan L. Ward
James T. Anderson
J. Todd Petty
Division of Forestry
West Virginia University
P. O. Box 6125
Morgantown, WV 26506
103
00/Month/0000
Ryan L. Ward
Division of Forestry
West Virginia University
P. O. Box 6125
Morgantown, WV 26506
Effects of road crossings on stream and stream-side salamander diversity, richness, and
abundance
Ryan L. Ward, Division of Forestry, West Virginia University, P.O. Box 6125, Morgantown,
WV 26506
James T. Anderson, Division of Forestry, West Virginia University, P. O. Box 6125,
Morgantown, WV 26506
J. Todd Petty, Division of Forestry, West Virginia University, P. O. Box 6125, Morgantown,
WV 26506
Abstract: Salamanders are important members of faunal communities in Appalachian streams,
and the use of salamanders as bioindicators is increasing. Roads are a necessary part of the
landscape, and have wide ranging ecological effects. Little is known of the effects of roads on
salamander diversity and abundance. Streams were sampled using quadrats in the flowing
channel, dry channel, and stream bank along transects to assess salamander diversity, richness
and abundance. Leaf litter bag sampling was also used to assess stream diversity. Akaike’s
Information Criterion was used for model selection at reach and stream scales in West Virginia.
Salamander diversity and richness was affected by elevation, stream gradient, canopy cover, and
the presence of roads. Overall, habitat models suggest habitat quality is the most important
factor affecting salamander richness. The presence of roads, stream gradient, and elevation
received the most empirical support for species’ abundances. Roads benefited disturbance
Written in the style of The Wildlife Society Bulletin.
Ward et al.
104
tolerant species while negatively affecting other species. Impacts that roads and culverts have on
habitat should be considered during the planning process and addressed through mitigation
efforts.
Key words: stream salamanders, culverts, passage, roads, northern two-lined salamanders,
Eurycea bislineata, Appalachian seal salamanders, Desmognathus monticol, northern spring
salamanders, Gyrinophilus porphyriticus, mountain dusky salamanders, Desmognathus
ochrophaeus
___________________
Wildlife Society Bulletin 00(0):000-000
Salamanders are important members of faunal communities in the Appalachian
Mountains. However, concerns have risen over apparent world-wide declines in amphibians
(including salamanders) (Blaustein 1994). Moreover, the importance of salamanders has been
well documented (Burton and Likens 1975a). In the Hubbard Brook Experimental Forest in
New Hampshire, Burton and Likens (1975a) found 6.5% of the total salamander biomass to
consist of stream and stream-side salamanders. The biomass produced each year by a population
of northern dusky salamanders (Desmognathus fuscus) ranged from 0.097-0.32 g per m2 of
streambed (Spight 1967). Annual energy flow through the salamander population in the
Hubbard Brook ecosystem was estimated at 11,000 kcal/ha, roughly equal to 20% of the energy
flow through bird and mammal populations (Burton and Likens 1975b). Salamanders convert
60% of the ingested energy into new tissue, making salamanders a good source of energy for
predators (Burton and Likens 1975b). Many variables are important for stream salamanders
including physical habitat, water quality, and potential predators. Roads also have many effects
on streams and stream salamanders.
Ward et al.
105
Physical habitat is important for salamander populations. Stream gradient explained 7%
of the variation in northern two-lined larvae (Eurycea bislineata) abundance in New Hampshire
(Barr and Babbitt 2002). Stream gradient is tied to channel substrate, with low gradient streams
being more likely to accumulate fine sediments (Murphy et al. 1981, Corn and Bury 1989).
Southern two-lined salamanders (Eurycea cirrigera), northern two-lined salamanders, northern
spring salamanders (Gyrinophilus porphyriticus), and Pacific giant salamanders (Dicamptodon
ensatus) avoid embedded areas high in fine sediments or bare rock (Murphy et al. 1981, Hawkins
et al. 1983, Barr and Babbitt 2002, Lowe and Bolger 2002, Smith and Grossman 2003).
Elevation explained 20% of the variation in the abundances observed in northern twolined larvae in New Hampshire (Barr and Babbitt 2002). Northern dusky and Appalachian seal
salamanders (Desmognathus monticola) only occur at elevations up to 1,189 m and 1,362 m
respectively (Green and Pauley 1987). Northern spring salamanders occur at elevations up to
1,279 m (Green and Pauley 1987). Diversity decreases as elevation increases out of species’
ranges.
Assemblages of salamanders differ along the course of a stream. Small salamander
larvae are susceptible to drift and can be found in higher abundance in downstream areas from
upstream hatching areas (Johnson and Goldberg 1975, Bruce 1986). The headwaters of small
streams can also serve as source populations of salamanders (Resetarits 1995).
Corn and Bury (1989) found that amphibian species richness, density and biomass were
highest in streams in unlogged watersheds. Abundance of northern two-lined salamanders and
northern dusky salamanders has been found to vary with canopy cover (Bast and Maret 1998,
Barr and Babbitt 2002). Rocco and Brooks (2000) found that northern two-lined salamanders
were abundant in streams of watersheds with highly fragmented forest cover. In contrast,
Ward et al.
106
northern spring salamanders had a low occurrence in streams with highly fragmented forest
cover in the watershed (Rocco and Brooks 2000). Willson and Dorcas (2003) found that the
relative abundance of northern dusky and southern two-lined salamanders was inversely
proportional to the amount of disturbed habitat within the stream’s watershed.
Water quality also affects salamander populations. Water temperature explained 18% of
the variation in abundance of northern two-lined larvae in New Hampshire (Barr and Babbitt
2002). Variation in northern two-lined salamanders and northern spring salamanders has been
correlated to pH (Bast and Maret 1998, Barr and Babbitt 2002). Most studies documenting the
effects of pH on stream salamanders are conducted in streams affected by acid mine drainage
and show strong effects (Kucken et al. 1994, Rocco and Brooks 2000). Acidification from acid
rain has less of an effect (Barr and Babbitt 2002). Jung et al. (2000) found that northern twolined salamanders in Shenandoah National Park were not affected by acidification from acid rain,
and Mitchell (1999) found that salamander diversity in the park was not affected.
Predators such as brook trout (Salvelinus fontinalis) can have strong ecological
interactions with salamanders (Resetarits 1997). Total salamander abundance is lower in stream
reaches containing brook trout (Resetarits 1997). Both northern two-lined and northern spring
salamander show reduced abundances in the presence of brook trout (Lowe and Bolger 2002,
Barr and Babbitt 2002). Survival of northern spring salamanders is reduced 35-50% in the
presence of brook trout (Resetarits 1991, Resetarits 1995).
Roads have many effects on streams and stream fauna. Roads are a source of pollution,
direct mortality, and sedimentation (Forman and Deblinger 2000). Road crossings can restrict
wildlife passage causing habitat fragmentation (Forman and Deblinger 2000). Upstream
movement of salamanders is important for the location of overwintering sites (Ashton and
Ward et al.
107
Ashton 1978, Ashton 1975). Roads can act as barriers to recolonization of disturbed areas (Jones
et al. 2000). Northern spring salamanders tended to disperse upstream (Lowe 2003), and
northern spring salamanders are more abundant in basins with paired streams, evidence of the
importance of dispersal and recolonization following disturbances (Lowe and Bolger 2002). Up
and downstream movements are important for the age structure of populations and a possible
means of density dependent regulation (Bruce 1986). Small isolated populations are vulnerable
to extinction from inbreeding depression, demographic, and environmental events (Lande 1993,
Mills and Smouse 1994).
Roads are a necessity for human society and constitute a major feature on the landscape.
Roads can have wide ranging ecological effects on the landscape. Forman and Deblinger (2000)
estimated an average width of 600 m for the zone of ecological impacts for a busy 4-lane
highway in Massachusetts. Forman (2000) extrapolated the ecological impacts of the highway to
determine that 1/5 of the land area in the United States was ecologically affected by public roads.
Angermeier et al. (2004) proposed examining the impacts of roads in 3 phases: road
construction, road presence, and urbanization. Road construction is often considered when
assessing environmental impacts, but road presence is often not considered and urbanization is
typically ignored (Angermeier et al. 2004).
The objectives of our study were to determine the habitat variables that best predict
salamander diversity and abundance and determine the effects of the presence of roads with
culverts on salamander communities. Because relationships become visible at different scales
across the landscape (Barr and Babbitt 2002), we looked for relationships at the stream and reach
scales.
Ward et al.
108
Study area
We conducted salamander sampling in the 11 digit hydrologic unit code (HUC)
watersheds of Glady Fork (1 culvert site, 1 reference site), Gandy Creek (4 culvert sites, 3
reference sites), and lower Dry Fork (4 culverts sites, 3 reference sites) in the Cheat River 8 digit
HUC (Figure 1) located in Randolph and Tucker County in eastern West Virginia (Seaber et al.
1987).
The average winter temperature in the Randolph and Tucker County was –0.5 ûC and the
average summer temperature was 20.1 ûC (Losche and Beverage 1967, Pyle et al. 1982).
Average annual rainfall for the two counties was 116 cm (Losche and Beverage 1967, Pyle et al.
1982). Prevailing winds in the study area occur from the northwest and west (Losche and
Beverage 1967, Pyle et al. 1982). Elevations in the study area ranged from 518 m to 1,390 m
(West Virginia Geographic Information Systems Technical Center 1999). Major geologic
groups were the Mauch Chunk, Chemung, Pottsville, Greenbrier, and Hampshire formation
(Cardwell et al. 1968). Soil associations included the Calvin-high base substratum-BelmontMeckesville association, Dekalb-Berks-Calvin association, Calvin association, Barbour-PopeSequatchie association, Dekalb-Gilpin association, and Dekalb-Calvin-Belmont association
(Losche and Beverage 1967, Pyle et al. 1982).
Methods
Our study was conducted from April—September 2004. We surveyed streams containing
a culvert (treatment streams) that had ≥150 m of length downstream before the mouth for
sampling. We also sampled reference streams without a culvert. We chose reference streams
based on geographic location with relation to sampled treatment streams. Streams with no
culverts that drained into the same main stem and were the same stream order (Strahler 1952)
Ward et al.
109
were used as the pool from which to randomly pick references. Treatment streams contained no
other culverts on the mainstem of the stream except for site 200 on an unnamed tributary to
Glady Fork. This site had an additional culvert near the mouth, but the culvert was far
downstream from the surveyed reaches (350 m) and was unlikely to affect populations further up
in the headwaters of the stream. The reference stream for this site also had a culvert at the mouth
of the stream, but far from the surveyed reaches (240 m).
Salamander sampling
We sampled transects for salamanders by sampling points every 30 m along the stream’s
thalweg. We used quadrat sampling due to its effectiveness in surveying stream-side
salamanders, dealing with habitat heterogeneity, and sampling different species with different
microhabitat requirements (Jaeger and Inger 1994). We surveyed 2 or 3 1 x 1 m quadrats at each
sampling point perpendicular to the stream. One quadrat was searched on the stream bank, 1 in
the dry substrate located in the stream channel (if present), and 1 in the substrate under the
flowing water of the stream (Figure 2). We used a coin flip to determine which stream bank was
surveyed at each sampling point (heads: stream right, tails: stream left).
We removed all cover objects from the quadrat and captured salamanders using aquarium
dip nets. We used plastic (Zip-lock) bags to hold each specimen for processing once sampling
was complete at the sampling point. The bags were placed in the shade and contained some
moisture to prevent desiccation and over heating. We identified each salamander to species,
measured total length, weighed, and identified the lifestage (adult or larva). We replaced cover
objects after sampling and released salamanders at their site of capture. Animal handling
protocols were approved by the West Virginia University Animal Care and Use Committee (030910).
Ward et al.
110
We sampled 2 separate transects in each stream (Figure 3). The lower transect started
240 m downstream from the outlet of the culvert and came up to the glide of the outlet pool. If
240 m of stream were not present at a site, then the transect began at the confluence with the
main stem. This occurred at site 443, which only had 210 m in its lower transect. The upper
transect started at the inlet of the culvert and continued upstream for 240 m. We used the 30 m
distance between sampling points to prevent pre-sampling disturbance and ensure independent
samples at each sampling point. Sampling in the reference reaches consisted of 2 240-m
transects with sampling points every 30 m. Sampling transects were centered on the location in
the stream length that would isolate the same percentage of stream length as the culverted stream
of interest.
Habitat assessment
We collected habitat data at each sampling point along each transect, and we calculated
additional variables using geographic information systems (GIS). We later used habitat variables
in the models tested for each transect.
Stream Gradient. We used a 10 m digital elevation model (DEM) from the West
Virginia Geographic Information Systems Technical Center (1999) to calculate stream gradients
along transects. We divided the difference in elevation at the bottom and top of each transect by
the total length of the transect to determine gradients of stream reaches. We used the elevations
from the uppermost sampling point on a stream and the lowermost point to determine gradient at
the stream scale.
Water Quality. We measured water temperature at each sampling point along the transect
using an Oakton® Acorn™ Series pH 6 pH/mV/ûC meter. We used the average from all 9
Ward et al.
111
sampling points as the reach water temperature, and the average from all 18 points for the stream
water temperature.
We measured stream pH at the top and bottom of each transect using the Oakton®
Acorn™ Series pH 6 pH/mV/ûC meter. We used the average of the 2 points as the reach pH. If
the readings at the top and bottom of the transect differed by more than a unit of 1, then we took
additional readings at sampling points to better obtain the changes along the transect, and the
average of all readings was used for the reach pH. We used all readings in a stream (normally 4)
to calculate the stream pH. Due to malfunctioning equipment, we obtained pH at both site 96
transects using the average of 2 YSI, Inc.® 650 MDS™ water quality meter readings in each
transect. The 2 readings were conducted in April and September in conjunction with a fish study
(Poplar-Jeffers, I. O. in progress).
Canopy Cover. We measured canopy cover using a spherical densiometer. We took 4
readings at each sampling point, facing upstream, downstream, stream right and stream left. We
calculated canopy cover for each point and used the average for the reach or stream depending
on the analysis.
Predators. Brook trout were the main fish predators in study area streams. We
conducted fish sampling in each reach where salamanders were sampled, using single-pass
electro-shocking with a Smith-Root, Inc® LR-24 Electrofisher™ backpack shocker. We
sampled a 150 m reach centered in each salamander transect. We calculated brook trout density
for each reach and used it as a predator variable for transects. We used the average density for
the stream for the predator variable in stream models.
Elevation. We used a 10 m DEM (West Virginia Geographic Information Systems
Technical Center 1999) to determine the elevation at each sampling point. We averaged the
Ward et al.
112
elevations of each sampling point to determine a reach (9 points) and stream elevation (18
points).
Reach Position. We searched transects above and below culverts on treatment streams.
On reference streams, we searched transects above and below a 30 m stream segment in the
place of a culvert. We assigned a value of 1 to upstream transects and a value of 2 to
downstream transects. This variable was not used for whole stream analyses.
Road Presence. We assigned a value of 1 to transects above and below culverts and a
value of 0 to transects on reference streams. We assigned a value of 1 or 0 to streams based on
the presence or absence of a road crossing between the sampled transects.
Data analysis
We analyzed data at different spatial scales. Reach scale generally consisted of 9
sampling points along a 240-m transect. Stream scale consisted of 2 240-m transects totaling 18
sampling points. We analyzed models explaining diversity, species richness, abundance of
various species and life stages, and total salamander abundance. Adult abundance and intolerant
abundance were analyzed as well because they, along with species richness and total abundance,
constitute and index of biotic integrity recently created and then validated for the Mid-Atlantic
Highlands (Southerland et al. 2004, Rocco et al. 2004). Simpson’s index of diversity was used to
assess diversity (Simpson 1949).
We used Akaike’s Information Criterion (AIC) for model selection. Because the number
of reaches (n = 32) and streams (n = 16) sampled was small, we corrected Akaike’s Information
Criterion for small sample size (AICc) (Burnham and Anderson 2002). Akaike weights were
calculated based on AICc values and used to rank models and for inference on the importance of
variables. SAS version 9.1 was used to perform statistical analysis (SAS Institute, Inc. 2003).
Ward et al.
113
Models. We used a total of 10 different a priori models in the AIC analysis. Three
models formed the basis for the models tested. Then we tested these models in different
combinations and with additional variables based on previous literature and biological
assumptions. All models tested with AIC were developed prior to the examination of data.
We used a road effect model combining the variables of road presence and reach
position. This model looked for differences in upstream versus downstream reaches, above and
below roads and for differences in streams crossed by roads versus streams without road
crossings. We tested the road effect model alone and then combined it with all other models
because it was a main interest of this study. The road effect model in whole stream analyses was
just the road presence variable.
We used a stream habitat model that combined elevation and stream gradient. Because
not all species’ ranges extended into higher elevations (Green and Pauley 1987), this variable
was expected to be important in predicting richness and the densities of salamander species.
Stream gradient affects the substrate in the streambed (Murphy et al. 1981, Corn and Bury 1989)
and we used it as a variable of suitable substrate for salamander use. We used the habitat model
in most combinations because of its expected importance. Additional variables also were added
to the habitat model. We used a model with brook trout density to test the effects of predators
and a model using canopy cover to examine the importance of riparian vegetation. We also used
a water quality model combining the habitat model, water temperature, and pH. This model was
not expected to be very important, because all streams had relatively good water quality.
We used multiple linear regression models to examine species richness and diversity. To
examine species’ abundances, we transformed data to total salamanders per ha and used negative
binomial regression models. Log likelihood values were calculated in SAS version 9.1 (SAS
Ward et al.
114
Institute, Inc. 2003), which does not use a constant derived from the dependent variable. This
does not affect AICc values because they are on a relative scale, and comparisons of models were
not conducted among different probability distributions (Burnham and Anderson 2002). The
global model is theoretically the best fitting model, so we used a chi square goodness-of-fit test
to assess the model structure and fit of the data for each set of models (Burnham and Anderson
2002).
Leaf Litter Bag Sampling
We conducted larval salamander sampling with the use of leaf litter bags in 15 streams
(Pauley and Little 1998, Chalmers and Droege 2002). We constructed bags by cutting 2.5 cm
netting into 30 x 30 cm squares. Then we placed 2 thin rocks approximately 10 x 10 cm on the
square, and covered it with 1.5 L of leaf litter. We wrapped the netting around the material, tied
it at the top with a zip tie, and attached flagging for visibility. We placed a bag in the flowing
channel at each 30 m sampling point, and we checked the bags in 2 weeks by quickly picking
them up and shaking them over a white plastic tub. To test leaf litter bag results, we used the
same multiple linear regression models tested for whole stream analyses. After model selection
had occurred, we used a pairwise t-test to compare species richness results from leaf litter bag
sampling with quadrat sampling to determine its effectiveness.
Results
Transect sampling
Estimates for each variable used as a parameter in models were calculated for both the
stream and reach scales (Table 1). A total of 476 salamanders (267 adults, 203 larvae, 6 escapes)
were captured representing 6 different species including northern two-lined salamanders,
Appalachian seal salamanders, mountain dusky salamanders (Desmognathus ochrophaeus),
Ward et al.
115
northern spring salamanders, and northern dusky salamanders (Table 2; Appendix 3). Two slimy
salamanders (Plethodon glutinosus) were captured in a rotten log on the bank at site LDR2.
Because the slimy salamander is an upland species it was excluded from analyses, bringing the
total number of species down to 5.
Species diversity. At the reach scale the stream and riparian habitat model was selected
as the best predicting model for the Simpson’s index and carried nearly all the weight (wi = 0.94,
R2 = 0.40) (Table 3).
At the stream scale, the road effect model had the highest weight, but poorly explained
the data (wi = 0.43, R2 = 0.03) (Table 3). The stream habitat model (wi = 0.20, R2 = 0.14) and
the habitat and predator model (wi = 0.19, R2 = 0.35) both received empirical support.
Species richness. At the reach scale, the stream and riparian habitat model had the
highest Akaike weight (wi = 0.84, R2 = 0.41) (Table 3). The remaining models received little to
no empirical support. The next best model was the stream and riparian habitat model with the
road effect model (wi = 0.12, R2 = 0.45).
At the stream scale, the stream and riparian model had the highest Akaike weight (wi =
0.51, R2 = 0.64) (Table 3). The stream habitat model was the second best model (wi = 0.25. R2 =
0.48). The third best model was the stream and riparian habitat with the road effect model (wi =
0.10, R2 = 0.56).
Salamander abundance. The global model at the reach scale fit the model well (ĉ =
1.33). The stream habitat model had the highest weight for predicting total salamander
abundance at the reach scale (Table 4). The stream and riparian habitat model also showed
evidence. At the stream scale, the global model showed some signs of overdispersion (ĉ =
Ward et al.
116
1.85), but the degrees of freedom were low (d.f. = 8). The stream habitat model also had the
highest weight with little evidence for other models (Table 5).
A large number of captures (35 adults, 146 larvae) allowed for the selection of models
predicting total, adult, and larval abundances for northern two-lined salamanders. The global
models at the reach scale fit well for predicting the abundance of adults (ĉ = 0.82) and total
northern two-lined salamanders (ĉ = 0.83). The global model for predicting larvae abundance
showed signs of underdispersion (ĉ = 0.40). At the reach scale, the stream habitat model had the
highest weight for predicting total northern two-lined abundance, followed closely by the road
effect model (Table 4). The road effect model had the highest weight for the densities of adult
salamanders, and the stream habitat model showed some support. The stream habitat model had
the highest weight for abundances of larval salamanders at the reach scale and the road effect
model also showed evidence (Table 6). The global model for total abundance at the stream scale
fit well (ĉ = 1.01), while the global models for adults (ĉ = 0.44) and larvae (ĉ = 0.59) showed
signs of slight underdispersion. The road effect model had the highest weight for total adult and
larval abundance at the stream scale, followed by the stream habitat model (Table 5).
Analysis of Appalachian seal salamanders was restricted to the abundance of adults (126
captures). The global model at the reach scale showed underdispersion (ĉ = 0.25) while the
global model at the stream scale fit reasonably well (ĉ = 1.21). The stream habitat with the road
effect model had the highest weight for predicting adult Appalachian seal salamanders at the
reach scale (Table 4). Several other models showed evidence including the stream habitat
model, habitat and predator model, and the stream and riparian habitat model. At the stream
scale, the stream habitat with road effect model had the highest weight, followed by the stream
habitat model (Table 5).
Ward et al.
117
Analysis of northern spring salamanders was restricted to the abundance of larvae (46
captures). The global model showed underdispersion at the reach scale (ĉ = 0.24), but fit well at
the stream scale (ĉ = 1.12). The road effect model had the highest weight for predicting larval
northern spring salamanders at the reach scale (Table 4). The stream habitat model also showed
evidence at the reach scale. At the stream scale, the road effect model had the highest weight,
followed by the stream habitat model (Table 5).
Analysis of mountain dusky salamanders was restricted to the abundance of adults (102
captures). The global model showed underdispersion at the reach scale (ĉ = 0.33), and slight
underdispersion at the stream scale (ĉ = 0.57). The stream habitat, road effect, and habitat and
predators models had the highest weights for predicting adult mountain dusky salamanders at the
reach scale (Table 4). At the stream scale, the road effect model and the stream habitat model
had the highest weights (Table 5).
Analysis of total intolerant salamanders included all salamanders except northern twolined salamanders (Southerland et al. 2004). The global model showed slight underdispersion at
the reach scale (ĉ = 0.70) and overdispersion at the stream scale (ĉ = 1.85). The stream habitat,
stream and riparian habitat, and habitat and predators models had the highest weights for
predicting abundance at the reach scale (Table 4). At the stream scale, the stream habitat with
road effect model had the highest weight (Table 5).
The global model for adult salamander abundance fit well at the reach scale (ĉ = 0.78)
and showed overdispersion at the stream scale (ĉ = 2.06). The stream habitat and stream and
riparian habitat models carried the most weight for predicting adult abundance at the reach scale
(Table 4). The stream habitat, road effect, and stream habitat with road effect models carried the
most weight at the stream scale (Table 5).
Ward et al.
118
Leaf litter bag sampling
We captured 39 salamanders using the leaf litter bags, recording 5 species (Table 6;
Appendix 4). Salamanders were caught at 10 out of 15 streams sampled with leaf litter bags.
Out of 144 bags placed at sampling points only 36 yielded captures. Pairwise t-tests showed the
mean number of salamander species detected was significantly lower with leaf litter bags (x̄ =
1.27, SE = 0.33) compared to quadrat sampling (x̄ = 3.40, SE = 0.21) (t14 = 5.03, P ≤ 0.001).
Species richness. The stream and riparian model had the highest Akaike weight (wi =
0.45) and a R2 = 0.74 (Table 7). The stream habitat model was the second best model with an
Akaike weight of 0.25 and a R2 = 0.65.
Discussion
Salamander diversity and richness
Reach Scale. The selection of the stream and riparian habitat model for best predicting
the Simpson’s index of diversity and species richness shows that good habitat is necessary for
maintaining salamander diversity. The effects of elevation and stream gradient on diversity and
richness were expected. Several species ranges do not extend to higher elevations (Green and
Pauley 1987). The stream gradient effect shows the importance of channel substrate. Low
gradient streams are more likely to accumulate fine sediments (Murphy and Hall 1981, Corn and
Bury 1989), and stream salamanders are negatively associated with fine sediments (Murphy et al.
1981, Hawkins et al. 1983, Barr and Babbitt 2002, Lowe and Bolger 2002, Smith and Grossman
2003). A variety of cover objects allows for the coexistence of different sized species
(Southerland 1986). Stream and stream-side salamanders are negatively affected by the loss of
canopy (Corn and Bury 1989, Bast and Maret 1998, Barr and Babbitt 2002, Willson and Dorcas
Ward et al.
119
2003). Many reaches sampled had open canopies and lacked riparian vegetation and had fewer
species of salamanders.
The road effect model received no empirical support for predicting Simpson’s index of
diversity. The road effect model when combined with the stream and riparian habitat model
showed the second highest weight for predicting richness, but had much less support. Roads
likely impact salamander richness, but the additional variables in the model did not explain
enough variation to overcome the stream and riparian habitat models alone in the model selection
process. Negative effects of roads may appear in other models such as the stream and riparian
habitat model. If a road creates a break in the canopy, this might give more support to models
with canopy cover as a variable. Roads also may alter the affects of stream gradient by causing
increased sedimentation.
Stream Scale. The road effect model received the most support for predicting Simpson’s
index of diversity, but the low R2 value indicates that the data are described poorly by the model,
and its selection may be aided by the low number of variables in the model. Salamander
communities dominated by northern two-lined salamanders where characteristic of streams
crossed by roads. The stream and riparian habitat model and the habitat and predator models
both received empirical support in predicting Simpson’s index of diversity showing the
importance of habitat and predation to salamander communities.
The stream and riparian habitat model received the most support for predicting richness
at the stream scale just like the reach scale. This further emphasizes its importance. The results
discussed for the reach scale apply for the same reasons. The stream habitat model alone
received some support for predicting richness, showing that elevation and gradient can strongly
affect diversity.
Ward et al.
120
The third and fourth models were the stream and riparian habitat and the stream habitat
models coupled with the road effect model as a variable. These 2 models received some
empirical support. This may indicate a change in habitat on streams crossed by roads, affecting
diversity. This may not be evident at the reach scale because of the distribution of different
species along gradients. Above and below reaches might have the same species richness, but
different composition. If composition is different, then the species richness for the stream will
be higher. This may have occurred on reference sites, where better habitats allowed for more
species establishment. Road sites may be dominated by species considered more tolerant and
assemblages vary little in composition.
Salamander abundance
Elevation and stream gradient were the most important variables predicting salamander
abundance. The stream habitat model showed the highest weight for predicting total salamander
abundance, larval two-lined abundance, total two-lined abundance, larval northern spring
abundance, adult mountain dusky abundance, and adult salamander abundance at the reach scale
and stream scale. The model also explained adult Appalachian seal salamander abundance at the
stream scale and intolerant abundance at the reach scale. The stream habitat model had the
second highest weight for adult two-lined abundance at both the reach and stream scale, and
adult Appalachian seal abundance at the reach scale. Elevation is important because different
species’ ranges do not extend into higher elevations. Stream gradient shows the importance of
channel substrate on salamander populations (Murphy and Hall 1981, Hawkins et al. 1983,
Southerland 1986, Corn and Bury 1989, Barr and Babbitt 2002, Lowe and Bolger 2002, Smith
and Grossman 2003). The stream habitat model was the only model that received empirical
support for predicting total salamander abundance at the stream scale.
Ward et al.
121
The road effect model had the highest weight for predicting adult two-lined abundance at
the reach scale and adult, larval, and total two-lined abundance at the stream scale. The road
effect model also was the best model for predicting larval northern spring salamander abundance
at the reach scale. The road effect model had the second highest weight for larval two-lined
abundance and total two-lined abundance at the reach scale, adult mountain dusky salamander
abundance at both the reach and stream scale, larval northern spring salamander abundance at the
stream scale, and adult salamander abundance at the stream scale. The road effect model carried
little weight for predicting total salamander abundance although it was the second best model at
the stream scale. The road model when coupled with the stream habitat model was the best
model for predicting the abundance of intolerant salamanders at the stream scale. Roads
possibly unbalance salamander communities, reducing some species and allowing some to
flourish causing total salamander abundance to remain unaffected. Our data suggests that
increased numbers of two-lined salamanders at streams crossed by roads offsets the lower
abundance of other species. The road presence parameter in the road effect model was negative
for all species except for northern two-lined salamanders (Appendix 6). Northern two-lined
salamanders are abundant in disturbed habitats (Rocco and Brooks 2000). The habitat
degradation and barrier effects of roads and culverts benefit a generalist species to the detriment
of other salamander species such as northern spring salamanders. Northern spring salamanders
are more abundant in undisturbed habitats (Rocco and Brooks 2000). The disturbance to
salamander habitat by roads includes sedimentation, loss of canopy cover, increased pollution,
and loss of population connectivity (Chapter 1).
We encountered some problems with fitting the negative binomial probability distribution
to some of the data. Several data sets fit the distribution at one scale, but not the other, possibly
Ward et al.
122
indicating the importance of changes in species’ abundance along the continuum of the stream.
Underdispersion was the most common problem we faced. This occurred because of the large
number of sites where few or no individuals of a particular species were encountered. At the
reach scale a species may only be caught in either the up or downstream transect. Failing to
detect a species with a low population density was exaggerated in the data after the
transformation to salamanders per ha. Often, we only captured a few individuals at a site, and
then calculated densities, resulting in redundancy in the data causing the underdispersion. In
some model sets the selection of the road effect and stream habitat models may be affected by
the low numbers of parameters in the 2 models. Our study did indicate an effect of roads, but the
study was not sufficient to determine the exact causation of the effect such as habitat
fragmentation, sedimentation, loss of canopy, or pollution.
Leaf Litter Bag Sampling
High flow events caused leaf litter bags to be an ineffective method of sampling. Some
bags were completely lost after floods, while others were found a few to a few hundred meters
downstream. Often the leaf litter in the bags would disintegrate from flowing water, and after
the 2 week period, the netting and rocks are all that would be found. The rate of species
detection was significantly lower for leaf litter bags, when compared to quadrat sampling along
transects. Others have noted problems with litter bags (Barr and Babbitt 2001, Chalmers and
Droege 2002).
AIC model selection for litter bags did mimic the selections at the reach and stream scale
for quadrat sampling along transects. The stream and riparian habitat model was selected first,
adding more evidence to its importance.
Ward et al.
123
Conclusions and Management implications
Habitat and species’ range was found to be important factors determining stream and
stream-side salamander diversity, richness, and abundance. The number of species decreased
with increasing elevation as did the abundances of species whose upper elevational ranges were
approached and exceeded. High gradient streams provide good habitat to a variety of species
and provided the necessary requirements for strong populations of species needing abundant
cover objects and a variety of different sized cover objects. Riparian vegetation was important
for maintaining forest canopy over streams, which benefited salamander diversity and the
abundance of some species. When planning future road routes and replacement of stream
crossing structures on existing roads, transportation planners and managers should realize that
these small streams, though often lacking fish populations, are important ecosystems and that
salamanders play important roles in these systems. When planners and managers understand the
importance of these small streams, they can plan to use alternatives and actions that will reduce
and limit impacts to salamanders and other members of stream communities.
Some limited evidence was found for a negative effect of roads on salamander diversity
and richness at the stream scale. These differences are probably due to increased human
disturbance on streams with roads. At the reach and stream scales, model selection showed
evidence that roads affect abundance of different species. Northern two-lined salamanders, a
disturbance tolerant species increased while other more sensitive species decreased.
Mitigation opportunities along roads for herpetofauna and other wildlife have received
increased attention (Forman et al. 2003, Aresco 2005). The effects of roads on habitat should be
considered when assessing the impacts of roads on salamanders. Road planners should limit the
Ward et al.
124
sedimentation of streams by applying the appropriate road surface such as gravel or asphalt for
roads with larger volumes of traffic (Reid and Dunne 1984). Stream networks are often
expanded by roads when they impede the downhill flow and collect water in ditches (Jones et al.
2000). Road planners should include more drainage culverts to allow water to pass under the
road instead of collecting in ditches and running to the nearest stream. This will help minimize
the effects of roads on streams by filtering water through vegetation and not altering the
discharge a stream channel has been formed to handle. The filtering of runoff will help reduce
the amount of sediment and pollutants in the water. Good water quality is important for
maintaining diverse salamander communities with good age structures (Rocco and Brooks 2000).
Mitigation opportunities exist for the replacement of existing culverts that impede
salamander passage and negatively affect salamander habitat. Unnatural disruptions in the
bedload transport of streams by road construction or presence are likely to affect salamander
diversity and overall populations. Many culverts in our study lacked geomorphic stability
(White 2004). Proper sizing and placement of culverts that allow for natural stream function are
needed. If conditions inside culverts mimic conditions in the surrounding stream channel,
salamanders are likely to be able to successfully negotiate the culvert and it will not act as a
barrier and fragment populations (Chapter 2). The installation of culverts that exceed the
channel width will benefit not only stream salamanders but other wildlife as well (Forman et al.
2003).
Disruption to the plant community is common around road crossings (Mader 1984, Miller
et al. 1997). The road planning process should include measures to prevent impacts to streamside vegetation and overstory canopy, and mitigate for disturbances by replanting and
reestablishing native riparian plants.
Ward et al.
125
Further research is needed to fully understand the effects of roads on stream and streamside salamanders and to better prevent negative effects on stream ecosystems. Research on
salamanders is needed to better understand their physical abilities and needs to overcome barriers
to up and downstream movements. Stream ecosystems would benefit from ways to prevent or
minimize the effects of roads on the sediment supply and discharge of streams. Research is
needed on the types, amounts, and effects of pollution from roads and methods to minimize their
entrance into streams. Once impaired, stream ecosystems are hard to restore to pre-disturbance
conditions, so as researchers and managers we should strive to find ways to accomplish our
needs such as transportation, while preventing negative effects to streams and their flora and
fauna.
Acknowledgements
Financial support for the study was provided by the West Division of Highways and West
Virginia University Division of Forestry. Guidance was provided by J. Steven Kite, Michael
Strager, and Ronald H. Fortney. Field assistance was provided by Ira Poplar-Jeffers, Pat Kish,
Josh White, and Jared Gregory. This is scientific article No. XXXX of the West Virginia
University Agricultural and Forestry Experiment Station.
Literature cited
Angermeier, P. L., A. P. Wheeler, and A. E. Rosenberger. 2004. A conceptual framework for
assessing impacts of roads on aquatic biota. Fisheries 29:19-29.
Aresco, M. J. 2005. Mitigation measures to reduce highway mortality of turtles and other
herpetofauna at a north Florida lake. Journal of Wildlife Management 69:549-560.
Ashton, Jr., R. E. 1975. A study of movement, home range, and winter behavior of
Desmognathus fuscus (Rafinesque). Journal of Herpetology 9:85-91.
Ward et al.
126
Ashton, Jr., R. E. and P. S. Ashton. 1978. Movements and winter behavior of Eurycea
bislineata (Amphibia, Urodela, Plethodontidae). Journal of Herpetology 12:295-298.
Barr, G. E. and K. J. Babbitt. 2002. Effects of biotic and abiotic factors on the distribution and
abundance of larval two-lined salamanders (Eurycea bislineata) across spatial scales.
Oecologia 133:176-185.
Barr, G. E. and K. J. Babbitt. 2001. A comparison of 2 techniques to sample larval stream
salamanders. Wildlife Society Bulletin 29:1238-1242.
Bast, K. and T. J. Maret. 1998. Ecological correlates of stream salamander abundance and
distribution. Journal of the Pennsylvania Academy of Science 71:164 (abstract).
Blaustein, A. R. 1994. Chicken Little or Nero’s fiddle? A perspective on declining amphibian
populations. Herpetologica 50:85-97.
Bruce, R. C. 1986. Upstream and downstream movements of Eurycea bislineata and other
salamanders in a southern Appalachian stream. Herpetologica 42:149-155.
Burnham, K. P. and D. R. Anderson. 2002. Model selection and multimodel inference a
practical information-theoretic approach second edition. Springer-Verlag New York, Inc.
New York, NY, USA.
Burton, T. M. and G. E. Likens. 1975a. Salamander populatons and biomass in the Hubbard
Brook Experimental Forest, New Hampshire. Copeia 1975:541-546.
Burton, T. M. and G. E. Likens. 1975b. Energy flow and nutrient cycling in salamander
populations in the Hubbard Brook Experimental Forest, New Hampshire. Ecology
56:1068-1080.
Cardwell, D. H., R. B. Erwin, and H. P. Woodward. 1968. Geologic map of West Virginia.
West Virginia Geological and Economic Survey. 1:250,000 scale. West Virginia
Ward et al.
127
Department of Environmental Protection and United States Geologic Survey Water
Resources Division. 1998 (digitized). UTM Zone 17 NAD 27.
Chalmers, R. J. and S. Droege. 2002. Leaf litter bags as an index to populations of northern
two-lined salamanders (Eurycea bislineata). Wildlife Society Bulletin 30:71-74.
Corn, P. S. and R. B. Bury. 1989. Logging in western Oregon: responses of headwater habitats
and stream amphibians. Forest Ecology and Management 29:39-57.
Forman, R. T. 2000. Estimate of the area affected ecologically by the road system in the United
States. Conservation Biology 14:31-35.
Forman, R. T. and R. D. Deblinger. 2000. The ecological road-effect zone of a Massachusetts
(U.S.A.) suburban highway. Conservation Biology 14:36-46.
Forman, R. T., D. Sperling, J. A. Bissonette, A. P. Clevenger, C. D. Cutshall, V. H. Dale, L.
Fahrig, R. France, C. R. Goldman, K. Heanue, J. A. Jones, F. J. Swanson, T. Turrentine,
and T. C. Winter. 2003. Road Ecology Science and Solutions. Island Press, Washington
D.C., U.S.A.
Green N. B. and T. K. Pauley. 1987. Amphibians and reptiles in West Virginia.
University of Pittsburgh Press, Pittsburgh, PA, USA.
Hawkins, C. P., M. L. Murphy, N. H. Anderson, and M. A. Wilzbach. 1983. Density of fish and
salamanders in relation to riparian canopy and physical habitat in streams of the
northwestern United States. Canadian Journal of Fisheries and Aquatic Sciences
40:1173-1185.
Jaeger, R. G. and R. F. Inger. 1994. Quadrat sampling. Pages 97-102 in W. R. Heyer, M. A.
Donnelly, R. W. McDiarmid, L. C. Hayek, and M. S. Fuller, editors. Measuring and
Ward et al.
128
monitoring biological diversity: standard methods for amphibians. Smithsonian
Institution Press. Washington, D. C.
Johnson, J. E. and A. S. Goldberg. 1975. Movement of larval two lined salamanders (Eurycea
bislineata) in the Mill River, Massachusetts. Copeia 1975:588-589.
Jones, J. A., F. J. Swanson, B. C. Wemple, and K. U. Snyder. 2000. Effects of roads on
hydrology, geomorphology, and disturbance patches in stream networks. Conservation
Biology 14:76-85.
Jung, R. E., S. Droege, J. R. Sauer, and R. B. Landy. 2000. Evaluation of terrestrial and
streamside salamander monitoring techniques at Shenandoah National Park.
Environmental Monitoring and Assessment 63:65-79.
Kucken, D. J., J. S. Davis, J. W. Petranka, and C. K. Smith. 1994. Anakeesta stream
acidification and metal contamination: effects on a salamander community. Journal of
Environmental Quality 23:1311-1317.
Lande, R. 1993. Risks of population extinction from demographic and environmental stochasity
and random catastrophes. The American Naturalist 142:911-927.
Losche, C. K., and W. W. Beverage. 1967. Soil survey of Tucker County and part of northern
Randolph County, West Virginia. United States Department of Agriculture, Soil
Conservation Service and Forest Service.
Lowe, W. H. 2003. Linking dispersal to local population dynamics: a case study using a
headwater salamander system. Ecology 84:2145-2154.
Lowe, W. H. and D. T. Bolger. 2002. Local and landscape-scale predictors of salamanders
abundance in New Hampshire headwater streams. Conservation Biology 16:183-193.
Ward et al.
129
Mader, H. J. 1984. Animal habitat isolation by roads and agricultural fields. Biological
Conservation 29:81-96.
Miller, Jr., R. L., D. R. Dewalle, R. P. Brooks, and J. C. Finley. 1997. Long-term impacts of
forest road crossings of wetlands in Pennsylvania. Northern Journal of Applied Forestry
14:109-116.
Mills, L. S. and P. E. Smouse. 1994. Demographic consequences of inbreeding in remnant
populations. The American Naturalist 144:412-431.
Mitchell, J. C. 1999. Amphibian diversity in three montane streams with different levels of
acidity, Shenandoah National Park, Virginia. Banisteria 14:28-35.
Murphy, M. L. and J. D. Hall. 1981. Varied effects of clear-cut logging on predators and their
habitat in small streams of the Cascade Mountains, Oregon. Canadian Journal of
Fisheries and Aquatic Science 38:137-145.
Murphy, M. L., C. P. Hawkins, and N. H. Anderson. 1981. Effects of canopy modification and
accumulated sediment on stream communities. Transactions of the American Fisheries
Society 110:469-478.
Pauley, T. K. amd M. Little. 1998. A new technique to monitor larval and juvenile salamanders
in stream habitats. Banisteria 12:32-36.
Poplar-Jeffers, I. O. in progress. Culvert inventory and effects on fish communities in a Central
Appalachian watershed. M.S. Thesis, West Virginia University, Morgantown, WV.
Pyle, R. E., W. W. Beverage, T. Yoakum, D. P. Amick, W. F. Hatfield, and D. E. McKinney.
1982. Soil survey of Randolph County area, main part, West Virginia. United States
Department of Agriculture, Soil Conservation Service and Forest Service.
Ward et al.
130
Reid, L. M. and T. Dunne. 1984. Sediment production from forest road surfaces. Water
Resources Research 20:1753-1761.
Resetarits, Jr., W. J. 1997. Differences in an ensemble of streamside salamander
(Plethodontidae) above and below a barrier to brook trout. Amphibia-Reptilia 18:15-25.
Resetarits, Jr., W. J. 1995. Competitive asymmetry and coexistence in size-structured
populations of brook trout and spring salamanders. Oikos 73:188-198.
Resetarits, Jr., W. J. 1991. Ecological interactions among predators in experimental stream
communities. Ecology 72:1782-1793.
Rocco, G. L. and R. P. Brooks. 2000. Abundance and distribution of a stream plethodontid
salamander assemblage in 14 ecologically dissimilar watersheds in the Pennsylvania
central Appalachians final report. Penn State Cooperative Wetlands Center, Forest
Resources Laboratory, Pennsylvania State University, University Park, PA Report No.
2000-4.
Rocco, G. L., R. P. Brooks, and J. T. Hite. 2004. Stream plethodontid assemblage response
(SPAR) index: development, application, and verification in the MAHA final report.
Penn State Cooperative Wetlands Center, Penn State Institutes of the
Environmental/Department of Geography, Pennsylvania State University, University
Park, PA. Report No. 2004-1.
SAS Institute, Inc. 2003. SAS version 9.1. Cary, North Carolina, USA.
Seaber, P. R., F. P. Kapinos, and G. L. Knapp. 1987. Hydrologic units maps. U.S. Geological
Survey Water-Supply Paper 2294.
Simpson, E. H. 1949. Measure of diversity. Nature 163:688.
Ward et al.
131
Smith, S. and G. D. Grossman. 2003. Stream microhabitat use by larval southern two-lined
salamanders (Eurycea cirrigera) in the Georgia piedmont. Copeia 2003:531-543.
Spight, T. M. 1967. Population structure and biomass production by a stream salamander. The
American Midland Naturalist 78:437-447.
Southerland, M. T. 1986. The effects of variation in streamside habitats on the composition of
mountain salamander communities. Copeia 1986:731-741.
Southerland, M. T., R. E. Jung, D. P. Baxter, I. C. Chellman, G. Mercurio, J. H. Volstad. 2004.
Stream salamanders as indicators of stream quality in Maryland, USA. Applied
Herpetology 2:23-46.
Strahler, A. N. 1952. Dynamic basis of geomorphology. Geological Society of America Bulletin
63:923-938.
West Virginia Geographic Information System Technical Center. 1999. Digital elevation model
10-meter. West Virginia Geologic and Economic Survey. UTM Zone 17 NAD 27.
White, J. A. 2004. Geomorphic analysis of stream crossings in a portion of the upper Cheat
River basin. M.S. Thesis, West Virginia University, Morgantown, WV.
Willson, J. D. and M. E. Dorcas. 2003. Effects of habitat disturbance on stream salamanders:
implications for buffer zones and watershed management. Conservation Biology 17:763771.
pH
Water Temperature (ûC)
6.57
14.25
0.19
0.51
0.02
0.06
2
Brook Trout Density (no./m )
3.15
76.11
Canopy Cover (%)
1.43
31.24
4.86
6.87
0.00
11.16
0.83
630.33
8.39
20.64
0.29
91.62
42.92
1160.56
SE Minimun Maximum
10.95
828.51
Mean
Reach Scale
Stream Gradient (%)
Elevation (m)
Variable
watersheds, West Virginia, 2004.
6.56
14.25
0.06
76.11
11.33
828.50
Mean
0.26
0.71
0.02
4.09
1.86
41.34
5.23
6.96
0.00
24.61
2.65
669.00
8.28
19.12
0.26
89.14
31.43
1115.00
SE Minimun Maximum
Stream Scale
Table 1. Habitat variables used in models for salamander diversity and abundance for the Dry Fork, Gandy Creek, and Glady Fork
132
Mean
2.19
7.88
0.13
6.38
0.13
Species
Northern Two-lined
Appalachian Seal
Northern Spring
Mountain Dusky
Northern Dusky
SE
0.13
0.98
0.09
2.86
0.58
Adults
Age
0.13
0.13
2.88
0.44
9.13
Mean
SE
0.13
0.09
0.73
0.24
3.57
Larvae
0.22
2.67
1.89
1.89
6.67
Mean
Above
0.22
0.96
0.68
0.77
3.27
SE
0.00
3.22
0.78
2.11
7.56
Mean
Below
Location From Culvert
0.00
1.26
0.32
1.05
3.04
SE
0.44
5.89
2.67
4.11
13.89
Mean
SE
0.29
1.57
0.91
1.38
6.32
Present
Roads
0.00
7.29
3.43
13.71
8.00
SE
0.00
1.19
1.19
6.03
4.03
Absent
Mean
Table 2. Salamander captures along transects in the Dry Fork, Gandy Creek, and Glady Fork watersheds, West Virginia, 2004.
133
EL, SG, CA, RD, RP
EL, SG, PR, RD, RP
EL, SG, RD, RP
EL, SG, WT, PH
EL, SG, CA
EL, SG, PR
EL, SG
RD,RP
Stream and Riparian Habitat plus Road Effect
Habitat, Predators, and Road Effect
Stream Habitat and Road Effect
Habitat and Water Quality
Stream and Riparian Habitat
Habitat and Predators
Stream Habitat
Road Effect
0.03
0.03
4
4
5
5
0.40
0.17
6
6
7
7
8
10
K
b
0.23
0.08
0.19
0.40
0.25
0.50
R
2
-97.58
-97.82
-99.83
-110.01
-99.41
-93.39
-94.10
-103.89
-92.98
-98.18
AICc
Reach Scale
12.43
12.20
10.18
0.00
10.61
16.62
15.91
6.12
17.03
11.84
∆AICc
c
0.00
0.00
0.01
0.94
0.00
0.00
0.00
0.04
0.00
0.00
wi
d
4
3
0.03
5
5
7
5
6
6
7
9
K
0.14
0.35
0.28
0.26
0.22
0.37
0.30
0.26
0.39
R
2
Simpson's Index of Diversity
-64.20
-62.62
-62.58
-61.13
-48.50
-59.82
-57.81
-56.12
-48.50
-31.70
AICc
Stream Scale
0.00
1.57
1.61
3.06
15.70
4.38
6.38
8.08
15.70
32.50
∆AICc
0.43
0.20
0.19
0.09
0.00
0.05
0.02
0.01
0.00
0.00
wi
Number of variables in model in addition to intercept and variance.
Difference between AICc and the best approximating model.
Akaike weight.
b
c
d
(RP) (reach scale only).
Elevation (EL), stream gradient (SG), canopy cover (CA), brook trout density (PR), water temperature (WT), pH (PH), road presence (RD), reach position
EL, SG, WT, PH, RD, RP
Habitat, Water Quality, and Road Effect
a
EL, SG, CA, PR, WT, PH, RD, RP
Model Parameters
Global
Model Name
a
Information Criterion corrected for small sample size, are in bold.
Dry Fork, Gandy Creek, and Glady Fork watersheds, West Virginia, 2004. The best approximating models, selected using Akaike’s
Table 3. Linear regression models of Simpson’s index of diversity and species richness for salamander communities in streams in the
134
Kb
10
8
7
7
6
6
5
5
4
4
R2
0.56
0.29
0.45
0.26
0.23
0.21
0.41
0.22
0.18
0.05
Table 3. Extended.
5.20
0.59
1.66
-7.26
5.18
4.39
6.37
-3.33
8.58
1.90
AICc
Reach Scale
c
12.46
7.86
8.92
0.00
12.44
11.65
13.63
3.93
15.85
9.16
∆AICc
0.00
0.02
0.01
0.84
0.00
0.00
0.00
0.12
0.00
0.01
d
wi
R2
0.02
0.48
0.54
0.64
0.50
0.56
0.59
0.65
0.58
0.73
Species Richness
3
4
5
5
6
5
6
6
7
9
K
0.01
-6.47
-4.15
-7.91
2.53
-4.62
-0.41
-2.98
6.64
19.17
AICc
Stream Scale
7.92
1.44
3.76
0.00
10.44
3.29
7.50
4.93
14.55
27.08
∆AICc
0.01
0.25
0.08
0.51
0.00
0.10
0.01
0.04
0.00
0.00
wi
135
EL, SG, CA, PR, WT, PH, RD, RP
EL, SG, WT PH, RD, RP
EL, SG, CA, RD, RP
EL, SG, PR, RD, RP
EL, SG, WT, PH
EL, SG, RD, RP
EL, SG, CA
EL, SG, PR
EL, SG
RD, RP
Habitat, Water Quality, and Road Effect
Stream and Riparian Habitat plus Road Effect
Habitat, Predators, and Road Effect
Habitat and Water Quality
Stream Habitat and Road Effect
Stream and Riparian Habitat
Habitat and Predators
Stream Habitat
Road Effect
a
Model Parameters
Global
Model Name
Criterion corrected for small sample size, are in bold.
4
4
5
5
6
6
7
7
8
10
b
K
AICc
1.25 -2970288.84
1.07 -2970298.77
1.12 -2970295.97
1.22 -2970297.89
1.15 -2970293.19
1.13 -2970293.37
1.20 -2970289.89
1.29 -2970291.56
1.17 -2970287.64
1.33 -2970282.92
ĉ c
d
9.93
0.00
2.80
0.88
5.58
5.40
8.88
7.21
11.12
15.85
∆AICc
Adults and Larvae
Total Salamanders
e
0.00
0.48
0.12
0.31
0.03
0.03
0.01
0.01
0.00
0.00
wi
Gandy Creek, and Glady Fork watersheds, West Virginia, 2004. The best approximating models, selected using Akaike’s Information
Table 4. Logistic models with a negative binomial distribution explaining salamander abundance on a reach scale in the Dry Fork,
136
.
7.93
8.01
8.78
2.48
5.52
2.95
2.99
0.24
0.00
0.72 -174303.37
0.30 -174303.29
0.33 -174302.51
0.49 -174308.81
0.32 -174305.77
0.31 -174308.34
0.31 -174308.30
0.30 -174311.05
0.21 -174311.29
∆AICc
11.66
AICc
0.82 -174299.63
ĉ
Adults
Table 4. Extended.
0.36
0.32
0.08
0.08
0.02
0.11
0.00
0.01
0.01
0.00
wi
AICc
0.51 -856432.70
0.42 -856433.52
0.44 -856430.69
0.42 -856430.93
0.44 -856428.28
0.47 -856428.64
0.45 -856425.02
0.43 -856425.22
0.43 -856422.17
0.40 -856414.99
ĉ
0.81
0.00
2.83
2.59
5.24
4.87
8.49
8.29
11.35
18.53
∆AICc
Larvae
0.28
0.42
0.10
0.12
0.03
0.04
0.01
0.01
0.00
0.00
wi
Northern Two-lined Salamanders
0.78
0.80
0.82
0.83
0.82
0.89
0.86
0.78
0.88
0.83
ĉ
-1080537.06
-1080537.22
-1080534.72
-1080534.48
-1080532.96
-1080533.98
-1080529.67
-1080530.82
-1080527.51
-1080522.57
AICc
0.16
0.00
2.50
2.74
4.25
3.24
7.55
6.40
9.71
14.65
∆AICc
Adults and Larvae
0.32
0.35
0.10
0.09
0.04
0.07
0.01
0.01
0.00
0.00
wi
137
ĉ
5.05
0.00
1.77
1.88
0.81
11.11
0.84 -699538.98
0.67 -699544.03
0.84 -699542.26
0.77 -699542.15
0.99 -699543.22
0.25 -699532.92
0.00
0.21
0.12
0.13
0.31
0.03
0.09
0.20 -223378.29
0.23 -223377.82
0.22 -223375.02
0.27 -223375.92
0.21 -223372.49
0.38 -223376.14
0.24 -223369.29
0.27 -223370.20
0.00
0.46
3.27
2.37
5.80
2.15
9.00
8.09
8.94
0.36
0.29
0.07
0.11
0.02
0.12
0.00
0.01
0.00
0.02
wi
AICc
0.29 -547234.47
0.25 -547235.28
0.26 -547233.39
0.29 -547233.24
0.28 -547229.51
0.25 -547230.32
0.27 -547227.07
0.30 -547227.03
0.28 -547223.97
0.33 -547217.35
ĉ
0.81
0.00
1.89
2.04
5.77
4.96
8.21
8.25
11.31
17.93
∆AICc
0.26
0.39
0.15
0.14
0.02
0.03
0.01
0.01
0.00
0.00
wi
e
Akaike weight.
Difference between AICc and the best approximating model.
Chi square goodness-of-fit statistic divided by degrees of freedom.
d
c
Number of variables in model in addition to intercept and variance.
2.60
0.63 -699541.43
0.06
0.43 -223369.34
6.07
∆AICc
b
3.26
0.70 -699540.77
0.05
0.49 -223372.22
AICc
Elevation (EL), stream gradient (SG), canopy cover (CA), brook trout density (PR), water temperature (WT), pH (PH), road presence (RD), reach position (RP).
3.78
0.52 -699540.25
0.00
wi
Adults
Mountain Dusky Salamanders
a
10.45
∆AICc
0.45 -699533.58
AICc
Larvae
Adults
ĉ
Northern Spring Salamanders
Appalachian Seal Salamanders
Table 4. Extended.
138
wi
e
ĉ c
3.35
0.46
1.80
0.00
5.28
0.46 -1736974.80
0.67 -1736977.69
0.51 -1736976.35
0.47 -1736978.15
0.65 -1736972.87
0.03
0.36
0.15
0.29
0.07
0.04
0.83 -1606489.12
0.60 -1606494.43
0.64 -1606492.06
0.74 -1606493.75
0.65 -1606489.80
0.57 -1606491.01
0.69 -1606486.82
5.30
0.00
2.36
0.68
4.62
3.42
7.61
6.67
e
0.03
0.41
0.13
0.29
0.04
0.07
0.01
0.01
0.00
0.00
wi
e
Akaike weight.
Difference between AICc and the best approximating model.
Chi square goodness-of-fit statistic divided by degrees of freedom.
d
c
Number of variables in model in addition to intercept and variance.
4.33
0.43 -1736973.82
0.02
0.78 -1606487.76
9.13
15.19
d
b
6.03
0.50 -1736972.11
0.04
0.59 -1606485.30
0.78 -1606479.24
∆AICc
Elevation (EL), stream gradient (SG), canopy cover (CA), brook trout density (PR), water temperature (WT), pH (PH), road presence (RD), reach position (RP).
4.35
0.67 -1736973.80
0.00
0.00
AICc
a
9.56
d
0.41 -1736968.58
∆AICc
15.21
AICc
Adults
Adults and Larvae
0.70 -1736962.94
ĉ c
Adult Salamanders
Intollerant Salamanders
Table 4. Extended.
139
EL, SG, WT PH, RD
EL, SG, CA, RD
EL, SG, PR, RD
EL, SG, WT, PH
EL, SG, RD
EL, SG, CA
EL, SG, PR
EL, SG
RD
Habitat, Water Quality, and Road Effect
Stream and Riparian Habitat plus Road Effect
Habitat, Predators, and Road Effect
Habitat and Water Quality
Stream Habitat and Road Effect
Stream and Riparian Habitat
Habitat and Predators
Stream Habitat
Road Effect
a
EL, SG, CA, PR, WT, PH, RD
Model Parameters
Global
Model Name
Criterion corrected for small sample size, are in bold.
3
4
5
5
5
6
6
6
7
9
b
K
1.52
1.32
1.42
1.39
1.41
1.63
1.52
1.53
1.71
1.85
ĉ c
-1456788.57
-1456790.96
-1456786.60
-1456786.79
-1456787.13
-1456782.08
-1456781.89
-1456781.81
-1456776.97
-1456758.15
AICc
2.39
0.00
4.36
4.17
3.84
8.89
9.08
9.15
14.00
32.81
∆AICcd
Adults and Larvae
Total Salamanders
0.18
0.58
0.07
0.07
0.09
0.01
0.01
0.01
0.00
0.00
w ie
Gandy Creek, and Glady Fork watersheds, West Virginia, 2004. The best approximating models, selected using Akaike’s Information
Table 5. Logistic models with a negative binomial distribution explaining salamander abundance on a stream scale in the Dry Fork,
140
0.00
0.00
0.00
0.00
0.00
0.02
0.44 -84442.36 30.49
0.64 -84455.67 17.17
0.28 -84459.97 12.87
0.32 -84459.74 13.10
0.39 -84461.24 11.60
7.77
8.18
8.13
3.85
0.00
0.31 -84465.07
0.38 -84464.66
0.41 -84464.71
0.37 -84468.99
0.24 -84472.84
0.83
0.12
0.01
0.01
wi
AICc
∆AICc
ĉ
Adults
Table 5. Extended.
AICc
0.53 -424536.73
0.64 -424536.14
0.66 -424531.79
0.55 -424532.10
0.69 -424531.78
0.69 -424526.76
0.70 -424526.46
0.55 -424526.86
0.69 -424521.09
0.59 -424502.33
ĉ
0.00
0.59
4.94
4.63
4.95
9.97
10.27
9.87
15.64
34.40
∆AICc
Larvae
0.49
0.37
0.04
0.05
0.04
0.00
0.00
0.00
0.00
0.00
wi
Northern Two-lined Salamanders
AICc
1.11 -535287.12
1.01 -535284.87
1.13 -535280.73
1.06 -535280.54
1.06 -535280.91
1.23 -535276.85
1.19 -535275.65
1.05 -535275.84
1.15 -535271.55
1.01 -535254.27
ĉ
0.00
2.25
6.40
6.58
6.21
10.27
11.48
11.28
15.58
32.86
∆AICc
Adults and Larvae
0.69
0.22
0.03
0.03
0.03
0.00
0.00
0.00
0.00
0.00
wi
141
0.00
0.00
0.04
0.06
0.00
0.56
1.21 -334509.38 29.79
1.16 -334527.43 11.74
5.09
4.34
4.23
5.23
2.17
5.96
1.04 -334534.82
0.97 -334528.67 10.50
0.00
1.04 -334534.08
0.95 -334539.17
0.75 -334534.94
1.08 -334533.94
0.87 -334537.00
0.31 -334533.20
AICc
∆AICc
0.27 -255488.07
0.26 -255484.91
0.26 -255481.36
0.28 -255480.55
0.29 -255480.64
0.37 -255476.50
0.27 -255476.04
0.32 -255475.33
0.48 -255470.62
0.00
3.16
6.71
7.52
7.44
11.57
12.04
12.74
17.45
36.34
∆AICc
0.77
0.16
0.03
0.02
0.02
0.00
0.00
0.00
0.00
0.00
wi
e
Akaike weight.
Difference between AICc and the best approximating model.
Chi square goodness-of-fit statistic divided by degrees of freedom.
d
c
Number of variables in model in addition to intercept and variance.
0.78
0.13
0.02
0.03
0.02
0.01
AICc
0.57 -255451.73
ĉ
b
0.00
3.51
7.72
6.34
7.73
8.01
0.00
0.00
0.00
0.00
wi
Elevation (EL), stream gradient (SG), canopy cover (CA), brook trout density (PR), water temperature (WT), pH (PH), road presence (RD).
0.32 -108544.14
0.31 -108540.63
0.31 -108536.42
0.48 -108537.80
0.34 -108536.41
0.79 -108536.13
0.35 -108531.13 13.01
0.54 -108532.54 11.60
0.86 -108529.47 14.67
1.12 -108524.06 20.08
ĉ
Adults
Mountain Dusky Salamanders
a
0.03
0.19
0.04
0.07
wi
∆AICc
AICc
Larvae
Adults
ĉ
Northern Spring Salamanders
Appalachian Seal Salamanders
Table 5. Extended.
142
w ie
ĉ c
0.00
5.70
5.41
3.14
6.60
1.09 -837882.49
1.36 -837876.80
1.59 -837877.09
1.33 -837879.36
1.04 -837875.89
0.02
0.13
0.04
0.04
0.63
0.02
1.09
1.06
1.19
1.14
0.99
1.56
1.09
-775727.85
-775729.04
-775724.96
-775724.89
-775727.21
-775726.44
-775721.91
-775721.98
1.18
0.00
4.07
4.15
1.82
2.59
7.13
7.05
5.53
24.59
0.21
0.38
0.05
0.05
0.15
0.11
0.01
0.01
0.02
0.00
e
Akaike weight.
Difference between AICc and the best approximating model.
Chi square goodness-of-fit statistic divided by degrees of freedom.
d
c
Number of variables in model in addition to intercept and variance.
6.57
1.48 -837875.93
0.07
1.06
-775723.50
-775704.45
wie
b
4.48
1.25 -837878.01
0.04
1.64
2.06
∆AICcd
Elevation (EL), stream gradient (SG), canopy cover (CA), brook trout density (PR), water temperature (WT), pH (PH), road presence (RD).
5.32
1.19 -837877.17
0.01
0.00
AICc
a
9.00
1.41 -837873.50
∆AICcd
27.74
AICc
Adults
Adults and Larvae
1.85 -837854.75
ĉ c
Adult Salamanders
Intollerant Salamanders
Table 5. Extended.
143
Mean
0.00
0.33
0.00
0.00
0.00
Northern Two-lined
Appalachian Seal
Northern Spring
Mountain Dusky
Northern Dusky
SE
0.00
0.00
0.00
0.23
0.00
Adults
Species
2004.
Age
0.07
0.07
0.33
0.47
1.33
Mean
SE
0.07
0.07
0.21
0.26
0.41
Larvae
0.00
0.00
0.00
0.63
1.88
Mean
SE
0.00
0.00
0.00
0.42
0.64
Present
Roads
0.14
0.14
0.71
1.00
0.71
Mean
SE
0.14
0.14
0.42
0.49
0.42
Absent
Table 6. Salamander captures for streams using leaf litter bags in the Dry Fork, Gandy Creek, and Glady Fork watersheds, West Virginia,
144
EL, SG, CA, PR, WT, PH, RD
Global
0.76
0.66
9
7
3
38.73
20.25
13.73
12.81
36.18
17.70
11.18
10.25
9.23
0.00
0.00
0.00
0.00
0.00
0.02
Akaike weight.
EL, SG, WT, PH, RD
Habitat, Water Quality, and Road Effect
0.05
6
11.78
5.90
0.05
d
RD
Road Effect
0.66
6
8.45
4.38
0.08
Difference between AICc and the best approximating model.
EL, SG, WT, PH
Habitat and Water Quality
0.68
6
6.93
3.46
0.39
0.45
wid
c
EL, SG, PR, RD
Habitat, Predators, and Road Effect
0.74
5
6.01
0.30
0.00
∆AICcc
Number of variables in model in addition to intercept and variance.
EL, SG, CA, RD
Stream and Riparian Habitat plus Road Effect
0.66
5
2.85
2.55
AICc
b
EL, SG, RD
Stream Habitat and Road Effect
0.68
5
4
Kb
Elevation (EL), stream gradient (SG), canopy cover (CA), brook trout density (PR), water temperature (WT), pH (PH), road presence (RD).
EL, SG, PR
Habitat and Predators
0.74
0.65
R2
a
EL, SG, CA
Stream and Riparian Habitat
Model Parametersa
EL, SG
Model Name
Stream Habitat
sample size.
Glady Fork watersheds, West Virginia, 2004. Model rankings were based on Akaike’s Information Criterion corrected for small
Table 7. Linear regression models of salamander diversity from leaf litter bag sampling in streams in the Dry Fork, Gandy Creek, and
145
Ward et al.
146
Figure 1. The 11-digit hydrologic units of lower Dry Fork, Gandy Creek, and Glady Fork in the
8-digit Cheat River hydrologic unit located in West Virginia. Numbered sites are streams with
roads and alphanumeric sites are reference streams.
Ward et al.
147
Figure 2. Diagram of typical sampling point along a transect. Quadrats (1x1 m) were searched
on the (A) bank, in the (B) flow, and (C) dry channel (if present).
Ward et al.
148
Upstream
Transect
30 m
Culvert or
additional 30 m
segment
Downstream
Transect
Figure 3. Diagram of 2 transects sampled on each stream. Each transect had 9 sampling points
and was either separated by a culvert or a 30 m stream segment.
149
Chapter IV:
Conclusions and Management Implications for Roads and Stream
Salamanders
Ryan L. Ward
James T. Anderson
Division of Forestry
West Virginia University
P. O. Box 6125
Morgantown, WV 26506
150
Conclusions and management implications for roads and stream salamanders
Abstract
Roads are a necessity for humans, but often negatively affect wildlife. Salamanders are
important members of stream ecosystems. Culverts are commonly used when roads cross small
streams, and these culverts can act as a barrier to salamander movement. In the Dry Fork and
lower Shavers Fork watershed in eastern West Virginia, we found that culverts were often
barriers to salamanders and fragmented a significant portion of the landscape. Salamander
communities in streams crossed by roads showed an increase in a disturbance tolerant species to
the detriment of other species. To allow for salamander passage, culverts should be designed to
minimize outlet hang, and maintain connectivity of stream channel substrate. Flow conditions
inside culverts should mimic the conditions in the natural channel. To lessen impacts on
salamander communities, roads should be designed to prevent negative effects to streams such as
sedimentation, pollution, excessive disruption to the riparian zone, and barriers to movement.
Transportation planners must understand and consider the ecological importance of streams and
their communities when designing future roads and maintain existing roads.
Introduction
Transportation planning should include measures to minimize ecological impacts of
roads. Road crossing designs should minimize impacts to stream functions and biological
communities. To better provide for stream function, geomorphic processes should be
incorporated into culvert design to improve crossings (White 2004). The ecological effects of
roads reach far outside the roadway itself (Forman and Delinger 2000). The impacts of roads
should be examined in three phases: road construction, road presence, and urbanization
This chapter written in the style of The Proceedings of the West Virginia Academy of Sciences.
151
(Angermeier et al. 2004). Road construction is often considered when assessing environmental
impacts, but road presence is often not considered and urbanization is typically ignored
(Angermeier et al. 2004).
In the Appalachians, stream salamanders are important members of the faunal
community, and often they are top predators. Road networks can have significant effects on
salamander communities, including habitat alteration, population isolation, trophic level
alteration, and direct mortality of individuals (Chapter 1). Angermeier et al. (2004) hypothesized
that road density is correlated with an increase in predominance in species tolerant to silt, metals,
petroleum products and salt, and species that are good colonizers. Roads can degrade
salamander habitats by removing canopy cover, increasing channelization, increasing
sedimentation, and affecting water quality (Chapter 1). Roads serve as animal barriers because
they create breaks in the microclimate, create disturbance, have environmentally unstable
margins, and result in the death of individuals through direct mortality (Mader 1984). Culverts
are commonly used at small stream crossings, and the hydraulic forces associated with culverts
often exceed the ability of salamander to successfully pass through during periods of movement
(Chapter 2). Culverts that prevent up and downstream movements of salamanders affect the
structure of populations and the ability of individuals to locate wintering sites (Ashton 1975,
Ashton and Ashton 1978, Bruce 1986). When populations become isolated they become more
vulnerable to extinction from inbreeding depression, demographic events, and environmental
events (Mills and Smouse 1994). Negative effects of habitat fragmentation and population
isolation have been documented (Chapter 1). The affects of roads on other stream organisms
also will affect salamanders. Culverts that exclude fish from stream reaches may allow
salamanders to unnaturally flourish, avoiding the effects of exploitative competition, changes in
152
demographics, and changes in spatial distribution (Chapter 1). Mortality from roads has been
implicated in the reduction of some amphibian populations (Fahrig et al. 1995).
The objectives of our study were to determine the extent of salamander habitat
fragmentation on the landscape, determine the effects of roads on salamanders, find road
construction and management options to reduce the impacts of roads, and determine future
research needed.
Study Area and Methods
Our study area included the 10 digit hydrologic unit code (HUC) watersheds of Glady
Fork, Dry Fork, and Shavers Fork in the Cheat River 8 digit HUC (Figure 1) located in Randolph
and Tucker County in eastern West Virginia (Seaber et al. 1987). Culvert surveys were
conducted in the entire study area (Chapter 2). Salamander surveys took place in a subset of the
area (Chapter 3).
The average winter temperature in the study area was –0.5 ûC, the average summer
temperature was 20.1 ûC, and the average annual rainfall was 116 cm (Losche and Beverage
1967, Pyle et al. 1982). Elevations ranged from 518 m to 1,472 m (West Virginia Geographic
Information System Technical Center 1999). The most abundant geologic map units were the
Pottsville group, Mauch Chunk group, Hampshire formation, and Chemung group (Carwell et al.
1968). Major soil associations were the Dekalb-Buchanan association, Calvin-high base
substratum-Belmont-Meckesville association, Dekalb-Calvin-Belmont association, Gilpin
association, Barbour-Pope-Sequatchie association, Calvin association, Dekalb-Berks-Calvin
association, Dekalb-Gilpin association, very stony land-Ernest-Brinkerton- Leetonia association,
and the Very stony land-Dekalb association (Losche and Beverage 1967, Pyle et al. 1982).
153
This study area was selected for its representative nature of the Mid-Atlantic highlands
area. Conditions found in the study area are common throughout the eastern part of West
Virginia and north into Pennsylvania. Our goal in selecting this area was to determine conditions
that are common in the Mid-Atlantic highlands and develop solutions that will solve problems
throughout the region.
We conducted culvert surveys from June—October 2003 (Chapter 2). This part of our
study inventoried existing culverts, and their potential to act as barriers for the movement of
salamanders. We also surveyed crossings for geomorphic stability (White 2004). We conducted
salamander surveys from April—September 2004 (Chapter 3). This part of our study examined
salamander diversity, richness and abundance in streams with road crossings and stream lacking
road crossings. Stream reaches above and below culverts also were compared with up and
downstream reaches on streams without culverts.
Results
Culverts created many barriers on the landscape. Culverts were classified using outlet
hang and the presence of continuous substrate as complete barriers at 55.0%, partial barriers at
34.2%, and nonbarriers at 10.8% of culverts surveyed (Chapter 2). The cumulative effects of
culverts resulted in barriers that isolated 20.6% of the total stream length in the Dry Fork
watershed and 17.4% in the Shavers Fork watershed (Chapter 2). Outlet hang was positively
correlated with stream gradient and culvert slope, and culverts containing streambed substrate
occurred on lower gradient streams, had lower culvert slope, and had a greater width compared
to culverts lacking substrate (Chapter 2). Only 17.9 % of the sites visited contained continuous
substrate through the culvert while most upstream reaches (87.3 %) and downstream reaches
(85.6 %) had continuous substrate (Chapter 2).
154
Salamander diversity and richness was affected the most by elevation, stream gradient,
ands canopy cover (Chapter 3). Roads influenced diversity, and the abundances of several
species (Chapter 3). Roads benefited disturbance tolerant species while negatively affecting
other species (Chapter 3).
Conclusions
Culverts pose a significant problem to the upstream movement of salamanders in the
study area. Isolated populations become vulnerable to inbreeding depression and demographic
and environmental stochasticity (Mills and Smouse 1994). Connectivity of populations allows
for genetic interchange, movement in response to environmental changes, and recolonization of
locally extirpated populations (Beier and Loe 1992). Fragmented populations of salamanders
have different levels of genetic diversity compared to continuous populations (Gibbs 1998). In
the study area, 55.0% of culverts were complete barriers and 34.2% were partial barriers, and the
cumulative effect of these barriers fragmented 20.6% of the total stream length in the Dry Fork
watershed and 18.4% in the lower Shavers Fork watershed (Chapter 2).
Quality habitat is the most important factor affecting salamander diversity and richness
(Chapter 3). Elevation and stream gradient were important model parameters that affected
salamander communities. Riparian cover was important for maintaining diversity and richness.
Roads had some effect on diversity and richness, but they showed greater importance in
predicting the abundances of individual species (Chapter 3). Northern two-lined salamanders
(Eurycea bislineata) benefited from the presence of roads, but Appalachian seal (Desmognathus
monticola), mountain dusky (Desmognathus ochrophaeus), and northern spring (Gyrinophilus
porphyriticus) salamanders were negatively affected (Chapter 3). The northern two-lined
salamander is a disturbance tolerant species (Rocco and Brooks 2000). Roads lowered diversity
155
in salamander communities by increasing the number of northern two-lined salamanders, while
decreasing other species.
Management Implications
Humans are dependent on roads, and stream and river crossings are inevitable, requiring
bridges and culverts. Biological communities should not be sacrificed to solve economic
transportation development problems with road construction. Culverts should be properly
installed to not only convey water, but to also allow for the passage of salamanders and other
stream organisms.
Preventing outlet hang should be a priority on all culverts installed, since excessive outlet
hang was a common problem. When outlet hang cannot be prevented, other crossing structures
such as bridges or fords should be considered. If an alternative crossing will not work in a
situation, weirs can be used to adjust stream gradients at the inlet and outlets of culverts,
compensating for large drops and hydraulic forces (Lauman 1976, Taylor and Love 2003). If
water is pooled into the culvert outlet this would help salamanders enter culverts.
Low roughness inside culverts allows 2-4 times the flow to pass as an equal section of
natural channel with equal slope (Bell 1973). Increasing the roughness of culverts slows water
velocities and would promote the passage of aquatic organisms. The retention of bedload
material inside culverts allows the roughness of the inside of the culvert to simulate natural
stream conditions, and the streambed material creates variations in the flow velocity that allow
salamanders to move upstream (Jackson 2003). Installing culverts with proper dimensions and
the correct placement of culverts can reduce the tendency to create breaks in the channel
substrate. Culverts with substrate had a lower mean slope than culverts lacking substrate, due to
lower water velocity (Chapter 2). Sites with continuous substrate had a lower mean stream
156
gradient compared to sites without continuous substrate (Chapter 2). High gradient streams
require steeper culverts. In these streams a standard corrugated steel pipe culvert may not suffice
and additional modifications may be required to lower flow velocity and allow deposition. High
gradient streams often form a series of step pools and a crossing structure that mimicked this step
pool configuration might better promote the passage of aquatic organisms. Velocities have been
slowed with the addition of corrugations and baffles inside culverts, which increase roughness
(Taylor and Love 2003). Baffles can improve fish passage and show good durability (McClellan
1970, Blevins and Carlson 1988). Problems with baffles include high cost, difficulty in
fabrication, sedimentation, debris jams, icing, and increased turbulence through the culvert
(Blevins and Carlson 1988, Baker and Votapka 1990, Fitch 1995). The increased turbulence
created by baffles may be negative for salamanders. However, if baffles trap some sediment this
might provide a suitable surface for salamanders to use during passage.
Standard corrugated steel pipe culverts can be used in low gradient streams and not create
a break in channel substrate. Culverts with continuous substrate had larger widths than culverts
lacking continuous substrate (Chapter 2). A larger diameter or width prevents pooling at the
inlet of a culvert and subsequent deposition of bed material before entering the culvert (Sylte
2002). White (2004) found that 91% of aggraded reaches at culverts within the study area were
at least partially caused by low conveyance. Small diameters or widths also constrict the flow of
streams, which can cause increased water velocity (Sylte 2002). Wide culverts better allow for
the retention of bed material and benefit salamanders and other aquatic organisms and allow the
water passing through to better simulate conditions in the natural channel.
We observed corrugated steel pipe culverts coated with asphalt on the inside in our study
area. The insides of culverts are sometimes coated with asphalt coverings to prolong the life of
157
the culvert (Baker and Votapka 1990). These coatings smooth out corrugations, decreasing
roughness, and affecting small fish that depend on the corrugations as resting bays during
passage (Baker and Votapka 1990). This practice decreases the roughness of culverts and likely
reduces the ability of salamanders to pass through culverts.
Effects of roads on habitat should be considered when assessing the impacts of roads on
salamanders (Chapter 3). Road planners should limit the sedimentation of streams by applying
the appropriate road surface such as gravel or asphalt for roads with larger volumes of traffic
(Reid and Dunne 1984), and when designing roads on steep slopes, planners should include more
drainage culverts to allow water to pass under the road instead of collecting in ditches and
running to the nearest stream (Jones et al 2000). The installation of more drainage culverts will
minimize the collection water in ditches will help lessen the effects of roads on streams. Water
will move past the road without collecting in a channel thus allowing vegetation to filter the
water, reducing the amount of sediment and pollutants in the water and not altering the discharge
a stream channel has been formed to handle.
Riparian vegetation was important for maintaining salamander diversity and the
abundance of some species (Chapter 3). Disruption to the plant community is common around
road crossings (Mader 1984, Miller et al. 1997). The road planning process should include
measures to prevent impacts to stream-side vegetation and overstory canopy, and mitigate for
disturbances by replanting and reestablishing native riparian plants. Geomorphically stable
crossings that do not require channelization and armoring of streambanks will help reduce
disturbance to the riparian area and salamander habitat.
Transportation planners and managers need to realize the importance of small streams.
Small streams, though often lacking fish populations, are important ecosystems and salamanders
158
play important roles, often as top predators. When managers and planners take the ecological
importance of small streams into consideration, they can plan ways to lessen the impacts of
humans. By designing culverts that allow weak swimming organisms such as salamanders to
pass, planners will benefit a wide range of species with varying swimming abilities. By
preserving and lessening the impacts to salamander habitats in streams, many other stream
organisms will benefit and the integrity of the ecosystem can be maintained.
Future Research
Much research is needed in the area of road, stream and salamander interaction. Most
research on culverts and barriers has concentrated on fish, and the swimming abilities of
salamanders are relatively unknown compared to most fish species (Chapter 2). More data on
the abilities of salamanders would better allow for the analysis of barriers (Chapter 2).
Future research needed includes the effects of fragmentation on stream salamander
populations and population genetics, and the movement of stream salamanders. Detailed studies
are needed on distances moved, reasons for movement, and timing of movement for different life
stages of salamanders.
Research on culvert designs should include ways to increase roughness, prevent increased
water velocities, and accommodate passage of a wider range of aquatic organisms. When
designing new culverts, consideration should be given to cost and ease of installation and
development of methods to retrofit existing culverts. The design of future crossings should also
focus on geomorphic stability and minimize disturbance to the stream channel and banks.
Stream ecosystems would benefit from ways to prevent or minimize the effects of roads
on the sediment supply and discharge of streams. Research is needed on the types, amounts, and
effects of pollutants from roads and methods to minimize their entrance into streams.
159
Further research is needed to fully understand the effects of roads on stream and streamside salamanders and to better prevent negative effects on stream ecosystems. Research into the
ecology of streams is important to understand the importance that each species plays in the
ecosystem function, and how to best maintain a diversity of species and functions with mounting
pressure from human disturbance.
Mitigation Opportunities
Existing crossings provide many opportunities for mitigation with the retrofitting or
replacement of existing culverts to restore stream connectivity for salamanders and other
organisms. Because 89.2% of existing culverts in the lower Shavers Fork and Dry Fork
watersheds form a barrier to salamanders and fragment a significant part of each watershed,
West Virginia Division of Highways (WVDOH) should receive mitigation credit from resource
agencies for replacing culverts that prevent the movement of salamanders and other stream
organisms and replacing them with crossing structures designed to promote fish and wildlife
passage. Outlet hang height and the disruption of the stream bedload should be used to assess
salamander passage of existing culverts and identify crossings in need of replacement. New
crossing structures should allow for flow conditions that mimic the natural channel and do not
exceed the physical abilities of stream organisms. Minimizing outlet hang should be a priority
on all new structures. Our study does not clearly indicate the amount of credit that should be
given for the replacement of culverts for salamanders, because we did not identify habitat
fragmentation as the cause of the change in species composition of salamander communities.
However replacement with a structure that benefits a weak swimming organism such as stream
salamanders will benefit a variety of stream organisms with varying swimming abilities, and the
160
amount of credit received for such an action should be higher than what is considered for
salamanders alone.
Opportunities also exist for the restoration of riparian areas at crossings. On the sides of
steep slopes, roads often travel next to streams for short distances as they form a tight bend
through the stream valley at crossings. Resource agencies should allow the WVDOH to plant
native plants in these areas and receive mitigation credit for implementing these plantings.
Restoring riparian areas around crossings reduces disturbance to salamander habitat and lessens
the impacts on salamander diversity and richness.
Restoring the sediment and hydrologic budget that formed the stream channel geometry
prior to road construction is another possibility for mitigation. Designing roads that let water
move past them without collecting and the installation of more drainage culverts are needed to
restore these budgets. When stream habitats undergo fewer disturbances, salamander
communities may not exhibit some of the negative effect of the presence of roads.
During the construction of new roadways, onsite mitigation should include the
installation of appropriate crossing structures that not only allow conveyance of water but also
provide for geomorphic stability and ecological function. Efforts should limit sediment
production from construction activities and the future road surface. Proper replanting of riparian
areas should occur after disturbance.
Our study indicated that species composition of salamander communities changed in
streams with a road crossing. This change involved the increase of a disturbance tolerant species
and can be quantified with the use of an index of biotic integrity (IBI). Rocco et al. (2004) tested
different IBI’s developed for assessing streams based on salamander communities. The
monitoring of changes in these indices can be used to evaluate the success of mitigation efforts.
161
Maintaining a diverse salamander community with strong populations of each species should be
the goal of road and culvert designs and mitigation efforts. WVDOH should receive mitigation
credit for implementing stream and habitat improvements that limit the ecological impacts of
roads to salamanders and other members of stream communities.
Literature Cited
Angermeier, P. L., A. P. Wheeler, and A. E. Rosenberger. 2004. A conceptual framework for
assessing impacts of roads on aquatic biota. Fisheries 29:19-29.
Ashton, R. E. 1975. A study of movement, home range, and winter behavior of Desmognathus
fuscus (Rafinesque). Journal of Herpetology 9:85-91.
Ashton, R. E. and P. S. Ashton. 1978. Movements and winter behavior of Eurycea bislineata
(Amphibia, Urodela, Plethodontidae). Journal of Herpetology 12:295-298.
Baker, C. O. and F. E. Votapka. 1990. Fish passage through culverts. Report No. FHWA-FL90-006. United States Department of Agriculture, Forest Service Technology and
Development Center, San Dimas, CA.
Beier, P. and S. Loe. 1992. A checklist for evaluating impacts to wildlife movement corridors.
Wildlife Society Bulletin 20:434-440.
Bell, M. C. 1973. Fisheries handbook of engineering requirements and biological criteria.
Fisheries-Engineering Research Program, Corps of Engineers, North Pacific Division,
Portland, OR.
Blevins, V. and R. F. Carlson. 1988. Retrofit design of drainage structures for improved fish
passage: literature review. Report No. AK-RD-89-02. Water Research Center, Institute
of Northern Engineering, University of Alaska Fairbanks, Fairbanks, AK.
162
Bruce, R. C. 1986. Upstream and downstream movements of Eurycea bislineata and other
salamanders in a southern Appalachian steam. Herpetologica 42:149-155.
Cardwell, D. H., R. B. Erwin, and H. P. Woodward. 1968. Geologic map of West Virginia.
West Virginia Geological and Economic Survey. 1:250,000 scale. West Virginia
Department of Environmental Protection and United States Geologic Survey Water
Resources Division. 1998 (digitized). UTM Zone 17 NAD 27.
Fahrig, L., J. H. Pedlar, S. E. Pope, P. D. Taylor, and J. F. Wegner. 1995. Effect of road traffic
on amphibian density. Biological Conservation 73:177-182.
Fitch, G. M. 1995. Nonanadromous fish passage in highway culverts. Virginia Transportation
Research Council Report VTRC 96-R6.
Forman, R. T. and R. D. Deblinger. 2000. The ecological road-effect zone of a Massachusetts
(U.S.A.) suburban highway. Conservation Biology 14:36-46.
Gibbs, J. P. 1998. Genetic structure of redback salamander Plethodon cinereus populations in
continuous and fragmented forests. Biological Conservation 86:77-81.
Jackson, S. D. 2003. Ecological considerations in the design of river and stream crossings in C.
L. Irwin, P. Garrett, and K. P. McDermott, editors, 2003 Proceedings of the International
Conference on Ecology and Transportation. Center for Transportation and the
Environment, North Carolina State University, Raleigh, NC.
Jones, J. A., F. J. Swanson, B. C. Wemple, and K. U. Snyder. 2000. Effects of roads on
hydrology, geomorphology, and disturbance patches in stream networks. Conservation
Biology 14:76-85.
163
Lauman, J. E. 1976. Salmonid passage at stream-road crossings. A report with department
standards for passage of salmonids. Environmental Management Section, Department of
Fish and Wildlife, Portland, OR.
Losche, C. K., and W. W. Beverage. 1967. Soil survey of Tucker County and part of northern
Randolph County, West Virginia. United States Department of Agriculture, Soil
Conservation Service and Forest Service.
Mader, H. J. 1984. Animal habitat isolation by roads and agricultural fields. Biological
Conservation 29:81-96.
McClellan, T. J. 1970. Fish passage through highway culverts. United States Department of
Transportation, Federal Highway Administration and Oregon State Game Commission.
Portland, OR.
Miller, Jr., R. L., D. R. Dewalle, R. P. Brooks, and J. C. Finley. 1997. Long-term impacts of
forest road crossings of wetlands in Pennsylvania. Northern Journal of Applied Forestry
14:109-116.
Mills, L. S. and P. E. Smouse. 1994. Demographic consequences of inbreeding in remnant
populations. The American Naturalist 144:412-431.
Pyle, R. E., W. W. Beverage, T. Yoakum, D. P. Amick, W. F. Hatfield, and D. E. McKinney.
1982. Soil survey of Randolph County area, main part, West Virginia. United States
Department of Agriculture, Soil Conservation Service and Forest Service.
Reid, L. M. and T. Dunne. 1984. Sediment production from forest road surfaces. Water
Resources Research 20:1753-1761.
Rocco, G. L. and R. P. Brooks. 2000. Abundance and distribution of a stream plethodontid
salamander assemblage in 14 ecologically dissimilar watersheds in the Pennsylvania
164
central Appalachians final report. Penn State Cooperative Wetlands Center, Forest
Resources Laboratory, Pennsylvania State University, University Park, PA Report No.
2000-4.
Rocco, G. L., R. P. Brooks, and J. T. Hite. 2004. Stream plethodontid assemblage response
(SPAR) index: development, application, and verification in the MAHA final report.
Penn State Cooperative Wetlands Center, Penn State Institutes of the
Environmental/Department of Geography, Pennsylvania State University, University
Park, PA. Report No. 2004-1.
Seaber, P. R., F. P. Kapinos, and G. L. Knapp. 1987. Hydrologic units maps. U.S. Geological
Survey Water-Supply Paper 2294.
Sylte, T. L. 2002. Providing for stream function and aquatic organism passage: an
interdisciplinary design. Stream Notes January 2002:1-4.
Taylor, R. N. and M. Love. 2003. Fish passage evaluation at stream crossings. California
salmonid stream habitat restoration manual.
West Virginia Geographic Information System Technical Center. 1999. Digital elevation model
10-meter. West Virginia Geologic and Economic Survey. UTM Zone 17 NAD 27.
White, J. A. 2004. Geomorphic analysis of stream crossings in a portion of the upper Cheat
River basin. M.S. Thesis. West Virginia University, Morgantown, West Virginia, USA.
165
Figure 1. The Dry Fork, Glady Fork, and Shavers Fork 10 digit hydrologic code watersheds
located in the Cheat River 8 digit watershed in eastern West Virginia.
166
Appendices
167
Appendix 1. Results of filter classification of barrier types for stream salamander passage
through 120 culverts in the Dry Fork and lower Shavers Fork watersheds, West Virginia, 2003.
Site No.
83
96
97
98
106
107
109
114
116
121
123
130
295
297
316
319
320
323
334
343
344
360
363
378
379
394
408
418
419
442
443
452
85
187
209
299
353
366
377
a
Watershed
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
multiple barrel culverts
Outlet
Height Continuous
Hang (m) Substrate
0.77
N
1.20
Y
0.30
N
0.30
N
0.15
N
1.40
N
0.22
N
0.15
N
0.50
N
0.36
N
0.48
N
1.50
N
0.86
N
0.33
N
0.95
N
0.13
N
0.32
N
0.32
N
0.15
N
0.30
N
1.75
N
0.50
N
1.21
N
0.29
N
1.35
N
0.50
Y
0.22
N
0.33
N
1.10
N
0.40
N
0.10
N
0.20
N
0.00
Y
0.00
Y
0.00
Y
0.00
Y
0.00
Y
0.00
Y
0.00
Y
Barrier Type
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Nonbarrier
Nonbarrier
Nonbarrier
Nonbarrier
Nonbarrier
Nonbarrier
Nonbarrier
Fragmented
stream length (m)
310
3417
284
80
585
1496
803
1956
1358
243
2449
1364
509
634
583
1049
152
2231
936
378
500
107
297
2752
4186
3748
405
365
55
1715
3733
3270
3834
185
210
331
1074
1085
1329
168
Appendix 1. Continued.
Site No.
404
420
72
91
93
95
127
170
186
188
195
201
210
298
309
346
361
371
390
391
406
407
421
432
434
439
440
a
196
a
365
9
10
15
24
29
37
73
84
90
115
133
a
Watershed
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Dry Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
multiple barrel culverts
Outlet
Height Continuous
Hang (m) Substrate
0.00
Y
0.00
Y
0.00
N
0.07
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.05
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.00
N
0.05/0.00
N/N
0.00/0.00
N/N
0.69
N
0.50
N
0.23
N
0.30
Y
0.15
N
0.23
N
0.35
N
0.45
N
0.52
N
0.20
N
1.70
N
Barrier Type
Nonbarrier
Nonbarrier
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial/Partial
Partial/Partial
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Complete
Fragmented
stream length (m)
1041
787
83
331
1117
1131
168
2892
894
383
354
833
157
622
1209
2000
3354
886
681
302
561
959
937
1389
63
3958
338
742
18
254
567
112
142
940
985
1336
1338
24
1049
129
169
Appendix 1. Continued.
Site No.
135
136
137
138
139
140
142
143
144
193
194
206
237
238
239
240
244
252
264
265
266
267
369
45
100
101
262
14
28
43
44
58
86
102
132
134
251
268
293
a
218
a
260
a
Watershed
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
Shavers Fork
multiple barrel culverts
Outlet
Height Continuous
Hang (m) Substrate
Barrier Type
0.42
N
Complete
0.25
Y
Complete
0.34
N
Complete
1.60
N
Complete
0.55
N
Complete
0.23
N
Complete
0.52
N
Complete
0.20
N
Complete
0.13
N
Complete
0.25
Y
Complete
0.15
N
Complete
0.18
N
Complete
0.18
N
Complete
0.80
N
Complete
0.18
N
Complete
0.50
N
Complete
0.51
N
Complete
0.40
N
Complete
0.40
N
Complete
0.10
N
Complete
0.40
Y
Complete
1.10
N
Complete
0.15
N
Complete
0.00
Y
Nonbarrier
0.00
Y
Nonbarrier
0.00
Y
Nonbarrier
0.00
Y
Nonbarrier
0.00
N
Partial
0.06
N
Partial
0.00
N
Partial
0.00
N
Partial
0.08
N
Partial
0.03
N
Partial
0.08
N
Partial
0.00
N
Partial
0.00
N
Partial
0.00
N
Partial
0.00
N
Partial
0.00
N
Partial
0.07/1.07
N/N
Partial/Complete
0.00/0.00
N/N
Partial/Partial
Fragmented
stream length (m)
2024
914
3101
1024
1424
1342
1362
328
3192
833
300
446
247
335
345
1741
519
293
1499
18
1122
1081
1492
105
72
261
1339
485
2760
1400
523
181
308
4272
2203
1694
832
242
16
2739
1045
170
Appendix 2. Locations of fragmented stream segments in the Shavers Fork and Dry Fork
watersheds, West Virginia, 2003.
Stream reaches isolated by a complete barrier culvert in the Dry Fork watershed (excluding
Glady Fork, Laurel Fork, Red Creek, and Otter Creek).
171
Appendix 2. Continued.
Stream reaches isolated by a partial barrier culvert in the Dry Fork watershed (excluding Glady
Fork, Laurel Fork, Red Creek, and Otter Creek).
172
Appendix 2. Continued.
Stream reaches isolated by complete and partial barrier culverts in the Dry Fork watershed
(excluding Glady Fork, Laurel Fork, Red Creek, and Otter Creek).
173
Appendix 2. Continued.
Stream reaches isolated by a complete barrier culvert in the lower Shavers Fork watershed.
174
Appendix 2. Continued.
Stream reaches isolated by a partial barrier culvert in the lower Shavers Fork watershed.
175
Appendix 2. Continued.
Stream reaches isolated by complete and partial barrier culverts in the lower Shavers Fork
watershed.
176
Appendix 3. Salamander captures at each site along transects in the Dry Fork, Gandy Creek, and
Glady Fork watersheds, West Virginia, 2004.
Northern Two-lined
Appalachian Seal
Northern Spring
Mountain Dusky
Northern Dusky
Site
Adults
Larvae
Adults
Larvae
Adults
Larvae
Adults
Larvae
Adults
Larvae
Escapea
121
2
5
1
0
0
6
0
0
0
0
0
201
3
13
6
0
1
2
6
0
0
0
1
93
8
55
5
0
0
8
8
1
2
0
0
97
0
4
9
0
0
0
12
0
0
0
1
96
0
4
5
0
0
1
8
0
0
0
1
LDR1
0
0
3
3
0
1
8
0
0
0
0
LDR2
2
0
5
0
0
2
4
0
0
0
0
LDR3
2
0
13
2
0
0
11
1
0
0
0
443
3
4
0
0
0
3
0
0
0
0
1
452
5
0
0
0
0
2
2
0
0
0
0
UDR2
1
1
2
0
0
4
9
0
0
0
0
UDR3
5
8
0
0
0
6
9
0
0
0
0
GLR2
0
7
45
0
0
9
3
0
0
0
0
200
0
9
11
0
1
0
12
0
0
0
0
UDR1
1
29
21
2
0
2
6
0
0
0
2
432
4
9
0
0
0
0
4
0
0
0
0
Total
36
148
126
7
2
46
102
2
2
0
6
a
Individuals that escaped capture before identification and were used only in total salamander
abundance calculations.
177
Appendix 4. Salamander captures at each site using leaf litter bags in the Dry Fork watershed,
West Virginia, 2004.
Site
Appalachian Appalachian Northern Two- Northern Spring
Mountain
Northern
Seal Adults Seal Larvae lined Larvae
Larvae
Dusky Larvae Dusky Larvae
93
0
0
0
0
0
0
97
0
0
1
0
0
0
121
2
0
4
0
0
0
LDR1
0
2
0
1
0
1
LDR2
0
0
0
0
0
0
LDR3
0
2
1
1
1
0
201
0
3
5
0
0
0
432
0
0
2
0
0
0
443
0
0
1
0
0
0
452
0
0
0
0
0
0
UDR1
0
0
3
0
0
0
UDR2
0
0
0
0
0
0
UDR3
0
0
0
0
0
0
200
0
0
2
0
0
0
GLR1
3
0
1
3
0
0
Total
5
7
20
5
1
1
178
Appendix 5. Logistic regression models with a negative binomial distribution explaining the
abundance of salamanders in the Dry Fork, Gandy Creek, and Glady Fork watersheds, West
Virginia, 2003.
Total Salamander Abundance
Model Parametersa
Model Name
d.f.
X
2b
c
P- value K
AICc
∆AICcd wie
Reach Scale
Stream Habitat
EL, SG
29
31.08
0.362
4 -2970298.77
0.00 0.48
Stream and Riparian Habitat
EL, SG, CA
28
34.03
0.200
5 -2970297.89
0.88 0.31
Habitat and Predators
EL, SG, PR
28
31.42
0.299
5 -2970295.97
2.80 0.12
Habitat and Water Quality
EL, SG, WT, PH
27
30.55
0.290
6 -2970293.37
5.40 0.03
Stream Habitat and Road Effect
EL, SG, RD, RP
27
31.11
0.267
6 -2970293.19
5.58 0.03
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD, RP
26
33.53
0.147
7 -2970291.56
7.21 0.01
Habitat, Predators, and Road Effect
EL, SG, PR, RD, RP
26
31.23
0.220
7 -2970289.89
8.88 0.01
Road Effect
RD, RP
29
36.29
0.165
4 -2970288.84
9.93 0.00
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD, RP
25
29.21
0.255
8 -2970287.64
11.12 0.00
Global
EL, SG, CA, PR, WT, PH, RD, RP
23
30.58
0.133
10 -2970282.92
15.85 0.00
Stream Scale
Stream Habitat
EL, SG
13 17.13
0.193
4 -1456790.96
0.00
0.58
Road Effect
RD
14 21.26
0.095
3 -1456788.57
2.39
0.18
Stream Habitat and Road Effect
EL, SG, RD
12 16.97
0.151
5 -1456787.13
3.84
0.09
Stream and Riparian Habitat
EL, SG, CA
12 16.69
0.162
5 -1456786.79
4.17
0.07
Habitat and Predators
EL, SG, PR
12 17.08
0.147
5 -1456786.60
4.36
0.07
Habitat and Water Quality
EL, SG, WT, PH
11 17.97
0.082
6 -1456782.08
8.89
0.01
Habitat, Predators, and Road Effect
EL, SG, PR, RD
11 16.74
0.116
6 -1456781.89
9.08
0.01
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD
11 16.83
0.113
6 -1456781.81
9.15
0.01
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD
10 17.13
0.072
7 -1456776.97 14.00 0.00
Global
EL, SG, CA, PR, WT, PH, RD
8
0.064
9 -1456758.15 32.81 0.00
a
Elevation (EL), stream gradient (SG), canopy cover (CA), predator density (PR), water temperature (WT), pH
(PH), road presence (RD).
b
c
e
Chi square goodness-of fit statistic.
Number of parameters in model with intercept and variance.
d
14.78
Difference in AICc and best approximating model.
Akaike weight.
179
Appendix 5. Continued.
Northern two-lined salamanders (Eurycea bislineata) (reach scale)
Model Name
a
Model Parameters
2b
d.f.
X
29
6.19
c
P- value
K
0.999
4
AICc
∆AICc
-174311.29
0.00
d
wi
e
Adults
Road Effect
RD, RP
0.36
Stream Habitat
EL, SG
29
8.83
0.999
4
-174311.05
0.24
0.32
Habitat and Water Quality
EL, SG, WT, PH
27
13.20
0.988
6
-174308.81
2.48
0.11
Stream and Riparian Habitat
EL, SG, CA
28
8.65
0.999
5
-174308.34
2.95
0.08
Habitat and Predators
EL, SG, PR
28
8.80
0.999
5
-174308.30
2.99
0.08
Stream Habitat and Road Effect
EL, SG, RD, RP
27
8.58
0.999
6
-174305.77
5.52
0.02
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD, RP
25
18.06
0.840
8
-174303.37
7.93
0.01
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD, RP
26
7.88
0.999
7
-174303.29
8.01
0.01
Habitat, Predators, and Road Effect
EL, SG, PR, RD, RP
26
8.51
0.999
7
-174302.51
8.78
0.00
Global
EL, SG, CA, PR, WT, PH, RD, R 23
18.88
0.708
10
-174299.63
11.66
0.00
Stream Habitat
EL, SG
29
12.25
0.997
4
-856433.52
0.00
0.42
Road Effect
RD, RP
29
14.76
0.987
4
-856432.70
0.81
0.28
Habitat and Predators
EL, SG, PR
28
11.76
0.997
5
-856430.93
2.59
0.12
Stream and Riparian Habitat
EL, SG, CA
28
12.25
0.996
5
-856430.69
2.83
0.10
Habitat and Water Quality
EL, SG, WT, PH
27
12.58
0.992
6
-856428.64
4.87
0.04
Stream Habitat and Road Effect
EL, SG, RD, RP
27
11.80
0.995
6
-856428.28
5.24
0.03
Larvae
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD, RP
26
11.14
0.995
7
-856425.22
8.29
0.01
Habitat, Predators, and Road Effect
EL, SG, PR, RD, RP
26
11.83
0.992
7
-856425.02
8.49
0.01
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD, RP
25
10.81
0.994
8
-856422.17
11.35
0.00
Global
EL, SG, CA, PR, WT, PH, RD, R 23
9.26
0.995
10
-856414.99
18.53
0.00
Stream Habitat
EL, SG
29
23.28
0.764
4
-1080537.22
0.00
0.35
Road Effect
RD, RP
29
22.50
0.799
4
-1080537.06
0.16
0.32
Habitat and Predators
EL, SG, PR
28
22.88
0.739
5
-1080534.72
2.50
0.10
Total
Stream and Riparian Habitat
EL, SG, CA
28
23.19
0.723
5
-1080534.48
2.74
0.09
Habitat and Water Quality
EL, SG, WT, PH
27
24.10
0.625
6
-1080533.98
3.24
0.07
Stream Habitat and Road Effect
EL, SG, RD, RP
27
22.10
0.732
6
-1080532.96
4.25
0.04
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD, RP
26
20.30
0.777
7
-1080530.82
6.40
0.01
Habitat, Predators, and Road Effect
EL, SG, PR, RD, RP
26
22.25
0.675
7
-1080529.67
7.55
0.01
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD, RP
25
21.95
0.639
8
-1080527.51
9.71
0.00
Global
EL, SG, CA, PR, WT, PH, RD, R 23
19.13
0.694
10
-1080522.57
14.65
0.00
a
Elevation (EL), stream gradient (SG), canopy cover (CA), predator density (PR), water temperature (WT), pH
(PH), road presence (RD), reach position (RP).
b
c
Number of parameters in model with intercept and variance.
d
e
Chi square goodness-of fit statistic.
Difference in AICc and best approximating model.
Akaike weight.
180
Appendix 5. Continued.
Northern two-lined salamanders (Eurycea bislineata) (stream scale)
Model Parametersa
Model Name
d.f.
X 2b
P- value
Kc
AICc
∆AIC cd
wie
0.83
Adults
Road Effect
RD
14
3.32
0.998
3
-84472.84
0.00
Stream Habitat
EL, SG
13
4.75
0.980
4
-84468.99
3.85
0.12
Stream Habitat and Road Effect
EL, SG, RD
12
3.73
0.988
5
-84465.07
7.77
0.02
Habitat and Predators
EL, SG, PR
12
4.93
0.960
5
-84464.71
8.13
0.01
0.01
Stream and Riparian Habitat
EL, SG, CA
12
4.53
0.972
5
-84464.66
8.18
Habitat and Water Quality
EL, SG, WT, PH
11
4.33
0.959
6
-84461.24
11.60
0.00
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD
11
3.08
0.990
6
-84459.97
12.87
0.00
0.00
Habitat, Predators, and Road Effect
EL, SG, PR, RD
11
3.57
0.981
6
-84459.74
13.10
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD
10
6.36
0.784
7
-84455.67
17.17
0.00
Global
EL, SG, CA, PR, WT, PH, RD
8
3.54
0.896
9
-84442.36
30.49
0.00
Road Effect
RD
14
7.42
0.917
3
-424536.73
0.00
0.49
Stream Habitat
EL, SG
13
8.26
0.826
4
-424536.14
0.59
0.37
Stream and Riparian Habitat
EL, SG, CA
12
6.64
0.880
5
-424532.10
4.63
0.05
Habitat and Predators
EL, SG, PR
12
7.88
0.794
5
-424531.79
4.94
0.04
Stream Habitat and Road Effect
EL, SG, RD
12
8.23
0.767
5
-424531.78
4.95
0.04
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD
11
6.09
0.867
6
-424526.86
9.87
0.00
Habitat and Water Quality
EL, SG, WT, PH
11
7.55
0.753
6
-424526.76
9.97
0.00
0.00
Larvae
Habitat, Predators, and Road Effect
EL, SG, PR, RD
11
7.68
0.742
6
-424526.46
10.27
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD
10
6.92
0.733
7
-424521.09
15.64
0.00
Global
EL, SG, CA, PR, WT, PH, RD
8
4.70
0.789
9
-424502.33
34.40
0.00
Road Effect
RD
14
15.57
0.340
3
-535287.12
0.00
0.69
Stream Habitat
EL, SG
13
13.07
0.442
4
-535284.87
2.25
0.22
Stream Habitat and Road Effect
EL, SG, RD
12
12.72
0.390
5
-535280.91
6.21
0.03
Total
Habitat and Predators
EL, SG, PR
12
13.61
0.326
5
-535280.73
6.40
0.03
Stream and Riparian Habitat
EL, SG, CA
12
12.72
0.390
5
-535280.54
6.58
0.03
Habitat and Water Quality
EL, SG, WT, PH
11
13.52
0.261
6
-535276.85
10.27
0.00
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD
11
11.51
0.402
6
-535275.84
11.28
0.00
Habitat, Predators, and Road Effect
EL, SG, PR, RD
11
13.14
0.284
6
-535275.65
11.48
0.00
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD
10
11.52
0.318
7
-535271.55
15.58
0.00
Global
EL, SG, CA, PR, WT, PH, RD
8
8.05
0.429
9
-535254.27
32.86
0.00
a
Elevation (EL), stream gradient (SG), canopy cover (CA), predator density (PR), water temperature (WT), pH
(PH), road presence (RD), reach position (RP).
b
c
Number of parameters in model with intercept and variance.
d
e
Chi square goodness-of fit statistic.
Difference in AICc and best approximating model.
Akaike weight.
181
Appendix 5. Continued.
Adult Appalachian Seal salamanders (Desmognathus monticola)
Model Parametersa
Model Name
d.f
X 2b
P- value Kc
AICc
∆AICcd wie
Reach Scale
Stream Habitat and Road Effect
EL, SG, RD, RP
27 18.21
0.897
6 -699544.03
0.00
0.31
Stream Habitat
EL, SG
29 28.79
0.476
4 -699543.22
0.81
0.21
Stream and Riparian Habitat
EL, SG, CA
28 23.39
0.713
5 -699542.26
1.77
0.13
Habitat and Predators
EL, SG, PR
28 21.46
0.806
5 -699542.15
1.88
0.12
Habitat, Predators, and Road Effect
EL, SG, PR, RD, RP
26 16.38
0.927
7 -699541.43
2.60
0.09
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD, RP
26 18.11
0.872
7 -699540.77
3.26
0.06
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD, RP
25 13.02
0.976
8 -699540.25
3.78
0.05
Habitat and Water Quality
EL, SG, WT, PH
27 22.76
0.698
6 -699538.98
5.05
0.03
Global
EL, SG, CA, PR, WT, PH, RD, RP
23 10.42
0.988
10 -699533.58 10.45
0.00
Road Effect
RD, RP
29
7.34
0.999
4 -699532.92 11.11
0.00
Stream Habitat and Road Effect
EL, SG, RD
12 11.35
0.4991
5 -334539.17
0.00
0.56
Stream Habitat
EL, SG
13 11.34
0.5826
4 -334537.00
2.17
0.19
Stream and Riparian Habitat
EL, SG, CA
12
9.03
0.7003
5 -334534.94
4.23
0.07
Habitat, Predators, and Road Effect
EL, SG, PR, RD
11 11.41
0.4095
6 -334534.82
4.34
0.06
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD
11 11.39
0.4111
6 -334534.08
5.09
0.04
Habitat and Predators
EL, SG, PR
12 12.93
0.3741
5 -334533.94
5.23
0.04
Road Effect
RD
14
4.39
0.9926
3 -334533.20
5.96
0.03
Habitat and Water Quality
EL, SG, WT, PH
11 10.65
0.4730
6 -334528.67 10.50
0.00
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD
10 11.56
0.3155
7 -334527.43 11.74
0.00
Global
EL, SG, CA, PR, WT, PH, RD
8
0.2889
9 -334509.38 29.79
0.00
Stream Scale
a
Elevation (EL), stream gradient (SG), canopy cover (CA), predator density (PR), water temperature (WT), pH
(PH), road presence (RD), reach position (RP).
b
c
Chi square goodness-of fit statistic.
Number of parameters in model with intercept and variance.
d
e
9.67
Difference in AICc and best approximating model.
Akaike weight.
182
Appendix 5. Continued.
Larval northern spring salamanders (Gyrinophilus porphyriticus)
Model Parametersa
Model Name
d.f.
X 2b
P- value Kc
AICc
∆AICcd wie
Reach Scale
Road Effect
RD, RP
29
5.86
0.999
4 -223378.29
0.00
0.36
Stream Habitat
EL, SG
29
6.68
0.999
4 -223377.82
0.46
0.29
Habitat and Water Quality
EL, SG, WT, PH
27 10.36
0.998
6 -223376.14
2.15
0.12
Stream and Riparian Habitat
EL, SG, CA
28
7.61
0.999
5 -223375.92
2.37
0.11
Habitat and Predators
EL, SG, PR
28
6.21
0.999
5 -223375.02
3.27
0.07
Stream Habitat and Road Effect
EL, SG, RD, RP
27
5.73
0.999
6 -223372.49
5.80
0.02
Global
EL, SG, CA, PR, WT, PH, RD, RP
23 11.27
0.980
10 -223372.22
6.07
0.02
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD, RP
26
6.96
0.999
7 -223370.20
8.09
0.01
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD, RP
25 10.77
0.994
8 -223369.34
8.94
0.00
Habitat, Predators, and Road Effect
EL, SG, PR, RD, RP
26
6.36
0.999
7 -223369.29
9.00
0.00
Road Effect
RD
14
4.43
0.992
3 -108544.14
0.00
0.78
Stream Habitat
EL, SG
13
4.07
0.990
4 -108540.63
3.51
0.13
Stream and Riparian Habitat
EL, SG, CA
12
5.78
0.927
5 -108537.80
6.34
0.03
Habitat and Predators
EL, SG, PR
12
3.74
0.988
5 -108536.42
7.72
0.02
Stream Habitat and Road Effect
EL, SG, RD
12
4.05
0.983
5 -108536.41
7.73
0.02
Habitat and Water Quality
EL, SG, WT, PH
11
8.73
0.647
6 -108536.13
8.01
0.01
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD
11
5.96
0.876
6 -108532.54 11.60 0.00
Habitat, Predators, and Road Effect
EL, SG, PR, RD
11
3.83
0.975
6 -108531.13 13.01 0.00
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD
10
8.57
0.573
7 -108529.47 14.67 0.00
Global
EL, SG, CA, PR, WT, PH, RD
8
8.93
0.348
9 -108524.06 20.08 0.00
Stream Scale
a
Elevation (EL), stream gradient (SG), canopy cover (CA), predator density (PR), water temperature (WT), pH
(PH), road presence (RD), reach position (RP).
b
c
Number of parameters in model with intercept and variance.
d
e
Chi square goodness-of fit statistic.
Difference in AICc and best approximating model.
Akaike weight.
183
Appendix 5. Continued.
Adult mountain dusky salamanders (Desmognathus ochrophaeus)
Model Parameters a
Model Name
2b
P- value Kc
d.f X
AICc
∆AIC cd wie
Reach Scale
Stream Habitat
EL, SG
29 7.36
0.999
4 -547235.28
0.00
0.39
Road Effect
RD, RP
29 8.47
0.999
4 -547234.47
0.81
0.26
Habitat and Predators
EL, SG, PR
28 7.31
0.999
5 -547233.39
1.89
0.15
Stream and Riparian Habitat
EL, SG, CA
28 8.22
0.999
5 -547233.24
2.04
0.14
Habitat and Water Quality
EL, SG, WT, PH
27 6.68
0.999
6 -547230.32
4.96
0.03
Stream Habitat and Road Effect
EL, SG, RD, RP
27 7.43
0.999
6 -547229.51
5.77
0.02
Habitat, Predators, and Road Effect
EL, SG, PR, RD, RP
26 7.11
0.999
7 -547227.07
8.21
0.01
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD, RP
26 7.71
0.999
7 -547227.03
8.25
0.01
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD, RP
25 7.05
0.999
8 -547223.97 11.31 0.00
Global
EL, SG, CA, PR, WT, PH, RD, RP
23 7.56
0.999
10 -547217.35 17.93 0.00
Road Effect
RD
14 3.77
0.997
3 -255488.07
0.00
0.77
Stream Habitat
EL, SG
13 3.40
0.996
4 -255484.91
3.16
0.16
Habitat and Predators
EL, SG, PR
12 3.10
0.995
5 -255481.36
6.71
0.03
Stream Habitat and Road Effect
EL, SG, RD
12 3.48
0.991
5 -255480.64
7.44
0.02
Stream and Riparian Habitat
EL, SG, CA
12 3.38
0.992
5 -255480.55
7.52
0.02
Habitat and Water Quality
EL, SG, WT, PH
11 4.10
0.967
6 -255476.50 11.57 0.00
Habitat, Predators, and Road Effect
EL, SG, PR, RD
11 3.02
0.990
6 -255476.04 12.04 0.00
Stream and Riparian Habitat plus Road Effect
EL, SG, CA, RD
11 3.55
0.981
6 -255475.33 12.74 0.00
Habitat, Water Quality, and Road Effect
EL, SG, WT, PH, RD
10 4.84
0.902
7 -255470.62 17.45 0.00
Global
EL, SG, CA, PR, WT, PH, RD
8 4.53
0.806
9 -255451.73 36.34 0.00
Stream Scale
a
Elevation (EL), stream gradient (SG), canopy cover (CA), predator density (PR), water temperature (WT), pH
(PH), road presence (RD), reach position (RP).
b
c
Number of parameters in model with intercept and variance.
d
e
Chi square goodness-of fit statistic.
Difference in AICc and best approximating model.
Akaike weight.
Global Model
0.7005
Reach
Stream
Species Richness
Species Richness
3.0791
1.0363
Simpsons Index of Diversity Stream
-0.0029
-0.0016
-0.0004
-0.0002
-0.0913
-0.0660
-0.0028
-0.0074
0.0284
0.0345
0.0003
0.0061
-0.2497
0.4460
0.4655
0.3062
pH
-0.0694 0.4632
-0.0951 0.4491
-0.0015 -0.0174
-0.0230 0.0153
Intercept Elevation Gradient Canopy Predator Water Temperature
0.4992
Scale
Simpsons Index of Diversity Reach
Measure
2004.
-0.7689
0.4535
0.0140
0.0172
-0.4460
-0.0103
Road Absence Reach Position
Appendix 6. Parameter estimates for liner regression models in the Dry Fork, Gandy Creek, Glady Fork watersheds, West Virginia,
184
-0.0001
-0.0002
-0.0022
-0.0036
0.6631
3.0660
5.3026
-0.0489
-0.0449
-0.0062
-0.0074
0.0214
0.0284
0.0027
0.0070
7.0051
5.3019
0.8812
0.7656
-0.0038
-0.0026
-0.0003
-0.0002
-0.0348
-0.0263
-0.0044
-0.0028
Intercept Elevation Gradient
Intercept Elevation Gradient Canopy
0.2105
Stream Habitat Model
Stream and Riparian Habitat Model
Appendix 6. Extended.
7.1642
5.4456
0.9194
0.8152
-0.0043
-0.0030
-0.0004
-0.0004
-0.0258
-0.0201
-0.0022
-0.0007
2.3929
2.3018
0.5747
0.7938
Intercept Elevation Gradient Predator
Habitat and Predators Model
185
-0.0001
-0.0003
-0.0020
-0.0037
0.6824
3.4097
5.8007
-0.0532
-0.0388
-0.0069
-0.0069
0.0185
0.0310
0.0022
0.0074
-0.1944
-0.1187
0.0344
-0.0243
-0.4055
-0.0159
7.6506
5.5065
0.8906
0.7486
-0.0004
-0.0030
-0.0004
-0.0004
-0.0402
-0.0230
-0.0037
-0.0022
1.7504
1.9869
0.5066
0.7258
-0.3792
0.4696
0.0402
0.0552
0.3955
0.0120
Intercept Elevation Gradient Predator Road Absence Reach Position
Intercept Elevation Gradient Canopy Road Absence Reach Position
0.2226
Habitat and Predators plus Road Effect Model
Stream and Riparian Habitat plus Road Effect Model
Appendix 6. Extended.
186
-0.0003
-0.0003
-0.0026
-0.0041
0.6732
0.8398
5.3001
7.6706
-0.0498
-0.0303
-0.0065
-0.0049
-0.4770
0.3448
0.0685
0.0862
-0.2792
0.0142
Intercept Elevation Gradient Road Absence Reach Position
Stream Habitat plus Road Effect Model
Appendix 6. Extended.
7.9636
6.3376
1.1692
1.4164
-0.0041
-0.0026
-0.0004
-0.0003
-0.0316
-0.0273
-0.0033
-0.0021
pH
0.0008 -0.1238
-0.0575 -0.0340
0.0018 -0.0398
-0.0190 -0.0534
Intercept Elevation Gradient Water Temperature
Habitat and Water Quality Model
187
-0.0003
-0.0003
-0.0024
-0.0039
1.2442
1.0956
4.7099
7.2945
-0.0634
-0.0463
-0.0044
-0.0042
-0.0453
-0.0966
0.0003
-0.0222
Intercept Elevation Gradient Water Temperature
pH
0.1727
0.2188
-0.0298
-0.0340
-0.7545
0.6940
0.0255
0.0582
-0.2855
0.0194
Road Absence Reach Position
Habitat and Water Quality plus Road Effect Model
Appendix 6. Extended.
3.5714
3.1181
0.5525
0.4584
Intercept
-0.2381
0.2222
0.0356
0.0651
-0.3750
0.0000
Road Absence Reach Position
Road Effect Model
188
Adults
Larvae
Adults
Appalachian Seal
Northern Spring
Mountain Dusky
-0.6516
1.2989
17.2495
2.6949
Adults
4.8651
Adults and Larvae
Northern two-lined
-9.2156
Adult Abundance
Adults
Northern two-lined
3.0839
6.5557
6.0117
Larvae
Northern two-lined
Global Model
-0.0012
-0.0021
-0.0002
0.0070
-0.0255
-0.0004
0.0156
-0.0020
-0.0012
0.0068
0.0038
0.0392
-0.3992
-0.1632
-0.2475
-0.2798
-0.2431
-0.0090
0.0173
0.0255
0.0167
0.1512
0.0131
0.0453
0.0075
0.0390
0.0171
0.4558
0.0101
3.7315
-22.8088
5.3663
-2.0736
-14.8815
-1.5198
-0.4478
pH
0.0665 0.3085
0.2796 0.5778
0.1452 0.6207
-1.0098 1.3286
0.0090 1.8401
-0.1654 1.0724
-0.8920 2.1866
-0.1342 1.1828
-0.0007 0.3044
Intercept Elevation Gradient Canopy Predator Water Temperature
Intollerant Abundance Adults and Larvae
Adults and Larvae
Age
Total Abundance
Species
Creek, Glady Fork watersheds, West Virginia, 2004.
-0.5199
-0.6490
-0.8207
-4.8335
-4.2999
-0.5481
-3.0293
-0.8525
-0.3420
0.0223
0.0945
0.8168
0.8351
1.0547
-0.7334
-0.9103
0.3358
0.0920
0.6746
0.9658
3.4679
4.2628
3.9331
1.9896
5.4097
4.5077
0.3534
Road Absence Reach Position Dispersion
Appendix 7. Parameter estimates for logistic models with a negative binomial distribution on a reach scale in the Dry Fork, Gandy
189
-0.0017
-0.0043
0.0007
-0.0029
-0.0210
-0.0015
-0.0001
-0.0019
-0.0013
11.9192
5.2315
10.6298
20.6094
4.6170
5.1831
7.0598
7.6389
0.0204
0.0292
0.0234
-0.0198
-0.0826
-0.0839
-0.0505
-0.0853
0.0100
0.0164
0.0242
0.0229
0.0477
0.0481
0.0051
0.0106
0.0000
0.0104
0.7491
1.0085
3.7043
6.7264
5.1815
2.3420
7.4712
4.8364
0.3864
9.3556
9.7545
7.4938
6.7268
24.8126
11.0569
5.9615
11.9188
10.1319
-0.0019
-0.0025
-0.0009
-0.0004
-0.0214
-0.0031
0.0006
-0.0043
-0.0021
0.0268
0.0346
0.0323
-0.0039
-0.0828
-0.0768
-0.0340
-0.0853
0.0158
0.7955
1.1382
3.7886
6.9376
5.4766
2.3472
7.5016
4.8364
0.0479
Intercept Elevation Gradient Dispersion
Intercept Elevation Gradient Canopy Dispersion
9.1021
Stream Habitat Model
Stream and Riparian Habitat Model
Appendix 7. Extended.
9.3677
9.8171
7.3874
6.8594
28.5185
10.7034
5.3429
11.5578
10.1437
-0.0021
-0.0029
-0.0014
-0.0008
-0.0269
-0.0025
0.0015
-0.0038
-0.0021
0.0328
0.0485
0.0541
0.0061
-0.0653
-0.0779
-0.0299
-0.0865
0.0167
1.1647
2.1346
3.3773
1.1898
7.3919
-1.8702
-2.0093
-2.1461
0.2187
0.7852
1.1058
3.6881
6.9322
5.2600
2.3277
7.4811
4.8041
0.4075
Intercept Elevation Gradient Predator Dispersion
Habitat and Predators Model
190
-0.0017
-0.0036
0.0005
-0.0023
-0.0249
-0.0016
0.0001
-0.0022
-0.0016
9.7536
3.5288
8.4429
27.0615
3.2609
4.4573
7.9398
8.2820
0.0165
0.0211
0.0222
-0.0215
-0.0795
-0.1400
-0.1564
-0.1335
0.0105
0.0134
0.0232
0.0299
0.0529
0.0083
0.0237
0.0365
0.0169
0.0107
-0.1889
-0.3111
0.2554
0.1792
-1.8272
1.0989
1.4578
0.9331
0.0233
-0.1135
0.0395
-0.1618
0.7441
0.8937
0.3356
0.2553
0.3494
0.0322
0.7412
0.9927
3.6882
6.5833
4.4612
2.1846
7.1395
4.7152
0.3861
9.8230
10.3591
7.6466
7.0527
27.8547
10.5381
6.9231
11.1711
10.2534
-0.0023
-0.0031
-0.0015
-0.0002
-0.0259
-0.0028
-0.0008
-0.0038
-0.0021
0.0244
0.0363
0.0505
-0.0508
-0.0443
-0.1005
-0.0874
-0.1028
0.0143
0.9679
1.6175
3.3825
-3.1379
4.3019
-0.4563
1.8343
-1.1583
0.0738
-0.3231
-0.5499
-0.0175
-0.4965
-1.8254
0.6807
0.9413
0.5586
-0.1180
-0.1411
0.0340
-0.1257
0.8250
0.7166
0.2852
0.1688
0.3851
0.0241
0.7611
1.0417
3.6838
6.7882
4.3929
2.2500
7.3333
4.7419
0.4054
Intercept Elevation Gradient Predator Road Absence Reach Position Dispersion
Intercept Elevation Gradient Canopy Road Absence Reach Position Dispersion
9.0104
Habitat and Predators plus Road Effect Model
Stream and Riparian Habitat plus Road Effect Model
Appendix 7. Extended.
191
-0.0021
-0.0042
0.0003
-0.0030
-0.0252
-0.0008
-0.0013
-0.0029
-0.0022
10.2530
11.4644
6.0712
10.3745
28.1108
7.0488
8.1024
10.4366
9.8677
0.0189
0.0255
0.0252
-0.0181
-0.0824
-0.1024
-0.0683
-0.1068
0.0140
-0.3471
-0.5978
-0.2138
-0.3307
-1.9621
0.7079
0.7942
0.6310
-0.1208
-0.1329
0.0694
-0.1691
0.6423
0.8230
0.2973
0.0824
0.3294
0.0250
0.7681
1.0602
3.7767
6.1829
4.4683
2.2509
7.3450
4.7497
0.4045
Intercept Elevation Gradient Road Absence Reach Position Dispersion
Stream Habitat plus Road Effect Model
Appendix 7. Extended.
7.8681
10.0726
4.3637
26.9529
30.9191
9.7260
7.2441
9.6965
9.4082
-0.0019
-0.0028
-0.0005
-0.0037
-0.0231
-0.0025
0.0021
-0.0038
-0.0020
0.0308
0.0428
0.0367
-0.0246
-0.0410
-0.1327
-0.1521
-0.1248
0.0161
0.4396
0.6354
0.4304
0.0412
pH
0.0493
0.1108 -0.0312
0.0734 -0.1872
0.1639
-0.6335 -1.3238
-0.0261 -0.7472
-0.1049
-0.4081
-0.0496
0.0282
Intercept Elevation Gradient Water Temperature
Habitat and Water Quality Model
0.7424
1.0897
3.6899
6.0232
5.2067
2.1923
6.5899
4.7001
0.4025
Dispersion
192
-0.0019
-0.0035
0.0060
-0.0023
-0.0260
-0.0030
-0.0005
-0.0027
-0.0018
8.8956
7.8906
3.4044
8.6477
18.7100
25.7377
2.4780
8.9651
7.3147
0.0164
0.0256
0.0257
-0.0482
-0.1989
-0.1542
-0.1371
-0.1585
0.0047
pH
0.0902 0.1526
0.0577 0.0705
0.1919 0.3505
-0.6872 -1.0070
0.0851 1.6332
-0.1290 0.6785
-0.6652 1.4553
-0.0940 0.8697
0.0098 0.1975
Intercept Elevation Gradient Water Temperature
-0.4968
-0.6500
-0.8465
-0.5862
-4.0953
-0.4540
-1.7647
-0.8270
-0.3788
-0.0004
0.1782
0.2015
-0.1802
1.3282
0.2726
-0.7938
0.3535
0.0654
0.7179
1.0399
3.6323
6.0015
4.0902
2.1675
6.2472
4.6427
0.3896
Road Absence Reach Position Dispersion
Habitat and Water Quality plus Road Effect Model
Appendix 7. Extended.
8.3727
8.4252
7.3199
6.1410
7.8400
7.2027
5.7994
6.8887
8.6818
Intercept
-0.3759
-0.5153
-0.1361
-0.2571
-0.9879
0.6986
0.7160
0.7403
-0.0650
-0.1090
0.0422
-0.1782
0.6103
-0.1574
0.1673
-0.2487
0.2840
0.0542
0.9197
1.3133
3.8764
6.8307
7.6970
2.3564
7.4381
4.9487
0.5343
Road Absence Reach Position Dispersion
Road Effect Model
193
Adults
Larvae
Adults
Appalachian Seal
Northern Spring
Mountain Dusky
-1.1207
-1.5648
20.9371
0.5252
Adults
5.9756
Adults and Larvae
Northern two-lined
-26.3245
Adults Abundnace
Adults
Northern two-lined
7.9474
6.6372
9.7926
Larvae
Northern two-lined
Global Model
-0.0016
-0.0036
-0.0002
0.0055
-0.0218
-0.0012
0.0213
-0.0069
-0.0015
0.0472
-0.0137
8.2283
-2.1316
0.0073
0.0083
-0.0504
1.4202
3.5701
0.1859 -22.8223
0.0103 -0.0005
0.0198
-0.4647
-4.4872
-1.0677
0.1207 -22.2661
0.0494
0.0130
0.0791 -0.0378
-0.2914
-0.3652
-0.5134
-0.0379
pH
0.0964 0.2919
0.0573 0.0885
0.1559 0.6868
-1.0084 1.0437
0.0198 0.4122
-0.2173 1.6010
-1.2826 3.8664
-0.2656 1.5312
0.0056 0.3594
Intercept Elevation Gradient Canopy Predator Water Temperature
Intollerants Abundance Adults and Larvae
Adults and Larvae
Age
Total Abundance
Species
Creek, Glady Fork watersheds, West Virginia, 2004.
-0.7111
-0.7248
1.1229
4.4580
3.1859
1.7978
6.2596
2.5735
0.6190
0.1304
0.1341
1.4375
1.1399
1.6255
1.0425
2.5653
3.2173
0.2201
Road Absence Dispersion
Appendix 8. Parameter estimates for logistic models with a negative binomial distribution on a stream scale in the Dry Fork, Gandy
194
-0.0023
-0.0109
0.0012
-0.0038
-0.0163
-0.0040
-0.0010
-0.0027
-0.0021
20.0375
4.7905
11.8040
14.9980
4.9738
7.6504
9.1719
9.3050
0.0185
0.0038
0.0116
0.0162
0.0175
-0.0874
0.0150
0.0040
0.0119
0.0013
0.0720
0.0599
0.0035 -0.1139
0.0061 -0.0129
-0.4716
-0.0029
0.2579
0.2413
1.7298
3.2498
2.3197
1.4196
4.5075
3.6455
0.2675
9.6646
10.1505
7.7588
8.0217
19.0785
12.1462
5.3491
20.9509
10.5632
-0.0022
-0.0028
-0.0010
-0.0015
-0.0158
-0.0040
0.0010
-0.0109
-0.0024
0.0142
0.0240
0.0183
-0.0398
0.0363
-0.1109
-0.0086
-0.4303
0.0000
0.2611
0.2679
1.7302
3.5532
2.6057
1.4226
4.5169
0.6397
0.2704
Intercept Elevation Gradient Dispersion
Intercept Elevation Gradient Canopy Dispersion
10.2177
Stream Habitat Model
Stream and Riparian Habitat Model
Appendix 8. Extended.
9.7070
10.3574
7.8734
8.6540
21.2927
11.6638
4.7064
21.3614
10.5540
-0.0023
-0.0034
-0.0017
-0.0028
-0.0199
-0.0033
0.0019
-0.0113
-0.0023
0.0177
0.0350
0.0337
-0.0191
0.1007
-0.1085
-0.0052
-0.4362
-0.0004
0.7956
2.1276
3.1100
3.2581
6.5802
-1.8292
-2.2775
0.7817
-0.1012
0.2567
0.2374
1.6521
3.5203
2.4597
1.4059
4.4947
3.6949
0.2703
Intercept Elevation Gradient Predator Dispersion
Habitat and Predators Model
195
-0.0025
-0.0100
0.0023
-0.0028
-0.0206
-0.0039
-0.0017
-0.0036
-0.0026
18.8548
3.1151
10.4134
22.7304
4.1244
8.5892
11.3377
10.6633
0.0109
0.0179
0.0236
0.0012
0.0834
0.0008
-0.0016 -0.0029
-0.0021
0.0094 -0.0046
-0.0801
-0.0143 -0.0079
-0.0957
0.0067
-0.4540
-0.0080
-0.4840
-0.7072
0.3447
-0.3466
2.4380
-0.5884
-1.0182
-0.3902
0.1901
0.2240
0.1728
1.7191
3.2386
1.7635
1.3735
4.3323
3.6233
0.2621
10.3229
11.3144
7.8141
8.6919
21.8294
11.3845
5.1263
21.2472
10.7192
-0.0025
-0.0037
-0.0016
-0.0027
-0.0211
-0.0030
0.0013
-0.0112
-0.0025
-0.0003
0.0069
0.0367
-0.0344
0.0368
-0.0942
0.0110
-0.4307
-0.0107
0.2466
1.1704
3.2494
2.2883
4.1251
-1.0847
0.7594
1.0351
-0.4852
-0.4283
-0.6477
-0.0733
0.2428
2.0720
-0.3569
-0.8106
-0.1054
0.2366
0.2250
0.1644
1.6513
3.5113
1.7173
1.3878
4.3939
3.6929
0.2610
Intercept Elevation Gradient Predator Road Absence Dispersion
Intercept Elevation Gradient Canopy Road Absence Dispersion
10.6067
Habitat and Predators plus Road Effect Model
Stream and Riparian Habitat plus Road Effect Model
Appendix 8. Extended.
196
-0.0026
-0.0108
0.0016
-0.0034
-0.0204
-0.0020
-0.0014
-0.0036
-0.0025
10.7269
20.8421
4.9073
11.6448
21.9916
8.4488
8.0646
11.4210
10.3422
-0.0019
-0.0021
0.0102
-0.0544
-0.0136
-0.0936
0.0109
-0.4271
-0.0078
-0.4410
-0.7191
0.2360
0.3916
2.3024
0.4179
-0.7524
-0.0446
0.2076
0.2254
0.1729
1.7218
3.5227
1.7694
1.3929
4.3959
3.5927
0.2623
Intercept Elevation Gradient Road Absence Dispersion
Stream Habitat plus Road Effect Model
Appendix 8. Extended.
8.0567
10.6799
4.2364
35.8665
29.1184
8.8717
4.9888
20.4454
9.6474
-0.0021
-0.0034
-0.0002
-0.0077
-0.0207
-0.0028
0.0040
-0.0099
-0.0023
0.0212
0.0311
0.0283
-0.1242
0.0288
-0.1317
-0.0009
-0.4280
0.0021
pH
0.1322 -0.0626
0.0962 -0.2365
0.1657 0.0478
-0.8003 1.6502
-0.0114 -0.9121
-0.0346 0.4540
-0.3672 0.4251
-0.1056 0.1708
0.0454 0.0308
Intercept Elevation Gradient Water Temperature
Habitat and Water Quality Model
0.1723
0.1858
1.6073
2.6445
2.1523
1.3025
3.9970
3.6286
0.2581
Dispersion
197
-0.0023
-0.0091
0.0072
-0.0027
-0.0199
-0.0079
-0.0002
-0.0034
-0.0020
8.5676
14.0957
-1.9767
6.0102
20.1550
36.5106
1.3597
10.0163
7.3235
-0.0065
0.0012
-0.0011
-0.1195
-0.0144
-0.2320
-0.1597
-0.5430
-0.0220
pH
0.1005 0.2030
0.0614 0.0499
0.1693 0.4516
-0.7934 -1.7323
0.0137 0.1742
-0.1681 1.2237
-0.7704 1.9992
-0.2735 1.4194
0.0093 0.2651
Intercept Elevation Gradient Water Temperature
-0.6878
-0.7460
1.0785
-0.1445
2.5489
1.5886
3.2198
2.5272
0.5888
0.1380
0.1445
1.5372
2.6462
1.7598
1.2113
3.7332
3.4387
0.2359
Road Absence Dispersion
Habitat and Water Quality plus Road Effect Model
Appendix 8. Extended.
8.3207
8.4484
7.0562
6.2651
6.7209
7.9753
6.3866
7.7469
8.6230
Intercept
-0.4132
-0.5578
0.1652
0.2050
1.0362
-0.6718
-0.6992
-0.6649
0.0886
0.3437
0.4000
1.7773
3.5785
4.0414
1.5270
4.4573
4.3415
0.3802
Road Absence Dispersion
Road Effect Model
198